|
Chapter 17. MySQL Cluster NDB 6.X/7.X
This chapter contains information about MySQL
Cluster, which is a high-availability, high-redundancy
version of MySQL adapted for the distributed computing environment.
Current releases of MySQL Cluster use versions 6 and 7 of the
NDBCLUSTER storage engine (also known
as NDB ) to enable running several
computers with MySQL servers and other software in a cluster.
Beginning with MySQL 5.1.24, support for the
NDBCLUSTER storage engine was removed
from the standard MySQL server binaries built by MySQL. Instead,
users of MySQL Cluster binaries built by MySQL should upgrade to the
most recent binary release of MySQL Cluster NDB 6.3 or MySQL Cluster
7.0 for supported platforms — these include RPMs that should
work with most Linux distributions. MySQL Cluster users who build
from source should be aware that, also beginning with MySQL 5.1.24,
NDBCLUSTER sources in the standard
MySQL 5.1 tree are no longer maintained; these users should use the
sources provided for MySQL Cluster NDB 6.2 or later. (Locations
where the sources can be obtained are listed later in this section.)
Note
MySQL Cluster NDB 6.1, 6.2, and 6.3 were formerly known as
“MySQL Cluster Carrier Grade Edition”. Beginning with
MySQL Cluster NDB 6.2.15 and MySQL Cluster NDB 6.3.14, this term
is no longer applied to the MySQL Cluster software — which
is now known simply as “MySQL Cluster” — but
rather to a commercial licensing and support package. You can
learn more about available options for commercial licensing of
MySQL Cluster from
MySQL
Cluster Features, on the MySQL web site.
This chapter contains information about MySQL Cluster in MySQL 5.1
mainline releases through MySQL 5.1.23, MySQL Cluster NDB 6.2
releases through 5.1.41-ndb-6.2.19, MySQL Cluster NDB 6.3
releases through 5.1.41-ndb-6.3.32, MySQL Cluster NDB 7.0
releases through 5.1.41-ndb-7.0.12 and MySQL Cluster NDB
7.1 releases through 5.1.41-ndb-7.1.2. Currently, the
MySQL Cluster NDB 6.3 and MySQL Cluster NDB 7.0 (formerly known as
“MySQL Cluster NDB 6.4”) release series are Generally
Available (GA). MySQL Cluster NDB 6.2, a previous GA release series,
is still supported, although we recommend that new deployments use
MySQL Cluster NDB 6.3 or MySQL Cluster NDB 7.0. MySQL Cluster NDB
7.1 is currently under development; we expect to make source and
binaries built from the MySQL Cluster NDB 7.1 available for
evaluation and testing purposes in the near future.
This chapter also contains historical information about MySQL
Cluster NDB 6.1, although this release series is no longer in active
development, and should not be used in new deployments. Users of
MySQL Cluster NDB 6.1 should upgrade to a later MySQL Cluster NDB
6.x or 7.x release series as soon as possible.
Platforms supported.
MySQL Cluster is currently available and supported on a number of
platforms, including Linux, Solaris, Mac OS X, and other
Unix-style operating systems on a variety of hardware. Beginning
with MySQL Cluster NDB 7.0, MySQL Cluster is also available (on an
experimental basis) on Microsoft Windows platforms. For exact
levels of support available for on specific combinations of
operating system versions, operating system distributions, and
hardware platforms, please refer to
http://www.mysql.com/support/supportedplatforms/cluster.html,
maintained by the MySQL Support Team on the MySQL web site.
We are continuing to work to make MySQL Cluster available on all
operating systems supported by MySQL; we will update the information
provided here as this work continues.
Availability.
MySQL Cluster NDB 6.2, MySQL Cluster NDB 6.3, and MySQL Cluster
NDB 7.0 binary and source packages are available for supported
platforms from http://dev.mysql.com/downloads/select.php?id=14.
Note
Binary releases and RPMs were not available for MySQL Cluster NDB
6.2 prior to MySQL Cluster NDB 6.2.15.
MySQL Cluster release numbers.
Starting with MySQL Cluster NDB 6.1 and 6.2, MySQL Cluster follows
a somewhat different release pattern from the mainline MySQL 5.1
Cluster series of releases. In this Manual and other MySQL
documentation, we identify these and later MySQL Cluster releases
employing a version number that begins with “NDB”.
This version number is that of the
NDBCLUSTER storage engine used, and
not of the MySQL server version on which the MySQL Cluster release
is based.
Version strings used in MySQL Cluster NDB 6.x and 7.x software.
The version string displayed by MySQL Cluster NDB 6.x and 7.x
software uses this format:
mysql-mysql_server_version -ndb-ndbcluster_engine_version
mysql_server_version represents the
version of the MySQL Server on which the MySQL Cluster release is
based. For all MySQL Cluster NDB 6.x and 7.x releases, this is
“5.1”.
ndbcluster_engine_version is the
version of the NDBCLUSTER storage
engine used by this release of the MySQL Cluster software. You can
see this format used in the mysql client, as
shown here:
shell> mysql
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 2
Server version: 5.1.41-ndb-7.0.12 Source distribution
Type 'help;' or '\h' for help. Type '\c' to clear the buffer.
mysql> SELECT VERSION()\G
*************************** 1. row ***************************
VERSION(): 5.1.41-ndb-7.0.12
1 row in set (0.00 sec)
This version string is also displayed in the output of the
SHOW command in the ndb_mgm
client:
ndb_mgm> SHOW
Connected to Management Server at: localhost:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=1 @10.0.10.6 (5.1.41-ndb-7.0.12, Nodegroup: 0, Master)
id=2 @10.0.10.8 (5.1.41-ndb-7.0.12, Nodegroup: 0)
[ndb_mgmd(MGM)] 1 node(s)
id=3 @10.0.10.2 (5.1.41-ndb-7.0.12)
[mysqld(API)] 2 node(s)
id=4 @10.0.10.10 (5.1.41-ndb-7.0.12)
id=5 (not connected, accepting connect from any host)
The version string identifies the mainline MySQL version from
which the MySQL Cluster release was branched and the version of
the NDBCLUSTER storage engine used.
For example, the full version string for MySQL Cluster NDB 7.0.5
(the first GA MySQL Cluster NDB 7.0 binary release) was
mysql-5.1.32-ndb-7.0.5 . From this we can
determine the following:
Since the portion of the version string preceding
“-ndb- ” is the base MySQL
Server version, this means that MySQL Cluster NDB 7.0.5
derives from the MySQL 5.1.32, and contains all feature
enhancements and bugfixes from MySQL 5.1 up to and including
MySQL 5.1.32.
Since the portion of the version string following
“-ndb- ” represents the
version number of the NDB (or
NDBCLUSTER ) storage engine,
MySQL Cluster NDB 7.0.5 uses version 7.0.5 of the
NDBCLUSTER storage engine.
New MySQL Cluster releases are numbered according to updates in the
NDB storage engine, and do not necessarily
correspond in a linear fashion with mainline MySQL Server releases.
For example, MySQL Cluster NDB 7.0.5 (as previously noted) is based
on MySQL 5.1.32, and MySQL Cluster NDB 7.0.6 is based on MySQL
5.1.34 (version string: mysql-5.1.34-ndb-7.0.6 ).
Compatibility with standard MySQL 5.1 releases.
While many standard MySQL schemas and applications can work using
MySQL Cluster, it is also true that unmodified applications and
database schemas may be slightly incompatible or have suboptimal
performance when run using MySQL Cluster (see
Section 17.1.5, “Known Limitations of MySQL Cluster”). Most of these issues
can be overcome, but this also means that you are very unlikely to
be able to switch an existing application datastore — that
currently uses, for example, MyISAM
or InnoDB — to use the
NDB storage engine without allowing
for the possibility of changes in schemas, queries, and
applications. Moreover, from MySQL 5.1.24 onwards, the MySQL
Server and MySQL Cluster codebases diverge considerably (and
NDB storage engine support dropped
from subsequent MySQL Server releases), so that the standard
mysqld cannot function as a dropin replacement
for the version of mysqld that is supplied with
MySQL Cluster.
MySQL Cluster development source trees.
MySQL Cluster development trees can also be accessed via
https://code.launchpad.net/~mysql/:
The MySQL Cluster development sources maintained at
https://code.launchpad.net/~mysql/ are licensed
under the GPL. For information about obtaining MySQL sources using
Bazaar and building them yourself, see
Section 2.3.3, “Installing from the Development Source Tree”.
Currently, MySQL Cluster NDB 6.2, MySQL Cluster NDB 6.3, and MySQL
Cluster NDB 7.0 releases are all Generally Available (GA), although
we recommend that new deployments use MySQL Cluster 6.3 or MySQL
Cluster 7.0. MySQL Cluster NDB 7.1 is in early development; we
intend to make the source tree for this release series available in
the near future. MySQL Cluster NDB 6.1 is no longer in active
development. For an overview of major features added in MySQL
Cluster NDB 6.x and 7.x, see
Section 17.1.4, “MySQL Cluster Development History”.
This chapter represents a work in progress, and its contents are
subject to revision as MySQL Cluster continues to evolve. Additional
information regarding MySQL Cluster can be found on the MySQL Web
site at http://www.mysql.com/products/cluster/.
Additional Resources.
More information may be found in the following places:
17.1. MySQL Cluster Overview
MySQL Cluster is a technology that enables
clustering of in-memory databases in a shared-nothing system. The
shared-nothing architecture allows the system to work with very
inexpensive hardware, and with a minimum of specific requirements
for hardware or software.
MySQL Cluster is designed not to have any single point of failure.
In a shared-nothing system, each component is expected to have its
own memory and disk, and the use of shared storage mechanisms such
as network shares, network file systems, and SANs is not recommended
or supported.
MySQL Cluster integrates the standard MySQL server with an in-memory
clustered storage engine called NDB
(which stands for “Network
DataBase”). In our
documentation, the term NDB refers to
the part of the setup that is specific to the storage engine,
whereas “MySQL Cluster” refers to the combination of
one or more MySQL servers with the NDB
storage engine.
A MySQL Cluster consists of a set of computers, known as
hosts, each running one or more processes.
These processes, known as nodes, may include
MySQL servers (for access to NDB data), data nodes (for storage of
the data), one or more management servers, and possibly other
specialized data access programs. The relationship of these
components in a MySQL Cluster is shown here:
All these programs work together to form a MySQL Cluster (see
Section 17.4, “MySQL Cluster Programs”. When data is stored by the
NDB storage engine, the tables (and
table data) are stored in the data nodes. Such tables are directly
accessible from all other MySQL servers (SQL nodes) in the cluster.
Thus, in a payroll application storing data in a cluster, if one
application updates the salary of an employee, all other MySQL
servers that query this data can see this change immediately.
Although a MySQL Cluster SQL node uses the mysqld
server damon, it differs in a number of critical respects from the
mysqld binary supplied with the MySQL
5.1 distributions, and the two versions of
mysqld are not interchangeable.
In addition, a MySQL server that is not connected to a MySQL Cluster
cannot use the NDB storage engine and
cannot access any MySQL Cluster data.
The data stored in the data nodes for MySQL Cluster can be mirrored;
the cluster can handle failures of individual data nodes with no
other impact than that a small number of transactions are aborted
due to losing the transaction state. Because transactional
applications are expected to handle transaction failure, this should
not be a source of problems.
Individual nodes can be stopped and restarted, and can then rejoin
the system (cluster). Rolling restarts (in which all nodes are
restarted in turn) are used in making configuration changes and
software upgrades (see
Section 17.2.6.1, “Performing a Rolling Restart of a MySQL Cluster”). In MySQL Cluster
NDB 7.0 and later, rolling restarts are also used as part of the
process of adding new data nodes online (see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”). For more
information about data nodes, how they are organized in a MySQL
Cluster, and how they handle and store MySQL Cluster data, see
Section 17.1.2, “MySQL Cluster Nodes, Node Groups, Replicas, and Partitions”.
Backing up and restoring MySQL Cluster databases can be done using
the NDB native functionality found in the MySQL Cluster management
client and the ndb_restore program included in
the MySQL Cluster distribution. For more information, see
Section 17.5.3, “Online Backup of MySQL Cluster”, and
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”. You can also
use the standard MySQL functionality provided for this purpose in
mysqldump and the MySQL server. See
Section 4.5.4, “mysqldump — A Database Backup Program”, for more information.
MySQL Cluster nodes can use a number of different transport
mechanisms for inter-node communications, including TCP/IP using
standard 100 Mbps or faster Ethernet hardware. It is also possible
to use the high-speed Scalable Coherent
Interface (SCI) protocol with MySQL Cluster, although
this is not required to use MySQL Cluster. SCI requires special
hardware and software; see
Section 17.3.5, “Using High-Speed Interconnects with MySQL Cluster”, for more about SCI
and using it with MySQL Cluster.
17.1.1. MySQL Cluster Core Concepts
NDBCLUSTER
(also known as NDB ) is an in-memory
storage engine offering high-availability and data-persistence
features.
The NDBCLUSTER storage engine can be
configured with a range of failover and load-balancing options,
but it is easiest to start with the storage engine at the cluster
level. MySQL Cluster's NDB storage
engine contains a complete set of data, dependent only on other
data within the cluster itself.
The “Cluster” portion of MySQL Cluster is configured
independently of the MySQL servers. In a MySQL Cluster, each part
of the cluster is considered to be a node.
Note
In many contexts, the term “node” is used to
indicate a computer, but when discussing MySQL Cluster it means
a process. It is possible to run multiple
nodes on a single computer; for a computer on which one or more
cluster nodes are being run we use the term cluster
host.
There are three types of cluster nodes, and in a minimal MySQL
Cluster configuration, there will be at least three nodes, one of
each of these types:
Management node (MGM node): The role of
this type of node is to manage the other nodes within the
MySQL Cluster, performing such functions as providing
configuration data, starting and stopping nodes, running
backup, and so forth. Because this node type manages the
configuration of the other nodes, a node of this type should
be started first, before any other node. An MGM node is
started with the command ndb_mgmd.
Data node: This type of node stores
cluster data. There are as many data nodes as there are
replicas, times the number of fragments (see
Section 17.1.2, “MySQL Cluster Nodes, Node Groups, Replicas, and Partitions”). For example,
with two replicas, each having two fragments, you need four
data nodes. One replica is sufficient for data storage, but
provides no redundancy; therefore, it is recommended to have 2
(or more) replicas to provide redundancy, and thus high
availability. A data node is started with the command
ndbd (see
Section 17.4.2, “ndbd — The MySQL Cluster Data Node Daemon”). In MySQL
Cluster NDB 7.0 and later, ndbmtd can also
be used for the data node process; see
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”, for more
information.
MySQL Cluster tables in MySQL 5.1 are normally
stored completely in memory rather than on disk (this is why
we refer to MySQL cluster as an
in-memory database). In MySQL 5.1,
MySQL Cluster NDB 6.X, and later, some MySQL Cluster data can
be stored on disk; see
Section 17.5.10, “MySQL Cluster Disk Data Tables”, for more
information.
SQL node: This is a node that accesses
the cluster data. In the case of MySQL Cluster, an SQL node is
a traditional MySQL server that uses the
NDBCLUSTER storage engine. An SQL
node is a mysqld process started with the
--ndbcluster and
--ndb-connectstring options, which are
explained elsewhere in this chapter, possibly with additional
MySQL server options as well.
An SQL node is actually just a specialized type of
API node, which designates any
application which accesses Cluster data. Another example of an
API node is the ndb_restore utility that is
used to restore a cluster backup. It is possible to write such
applications using the NDB API. For basic information about
the NDB API, see Getting Started with the NDB API.
Important
It is not realistic to expect to employ a three-node setup in a
production environment. Such a configuration provides no
redundancy; in order to benefit from MySQL Cluster's
high-availability features, you must use multiple data and SQL
nodes. The use of multiple management nodes is also highly
recommended.
For a brief introduction to the relationships between nodes, node
groups, replicas, and partitions in MySQL Cluster, see
Section 17.1.2, “MySQL Cluster Nodes, Node Groups, Replicas, and Partitions”.
Configuration of a cluster involves configuring each individual
node in the cluster and setting up individual communication links
between nodes. MySQL Cluster is currently designed with the
intention that data nodes are homogeneous in terms of processor
power, memory space, and bandwidth. In addition, to provide a
single point of configuration, all configuration data for the
cluster as a whole is located in one configuration file.
The management server (MGM node) manages the cluster configuration
file and the cluster log. Each node in the cluster retrieves the
configuration data from the management server, and so requires a
way to determine where the management server resides. When
interesting events occur in the data nodes, the nodes transfer
information about these events to the management server, which
then writes the information to the cluster log.
In addition, there can be any number of cluster client processes
or applications. These are of two types:
Standard MySQL clients.
MySQL Cluster can be used with existing MySQL applications
written in PHP, Perl, C, C++, Java, Python, Ruby, and so on.
Such client applications send SQL statements to and receive
responses from MySQL servers acting as MySQL Cluster SQL
nodes in much the same way that they interact with
standalone MySQL servers.
MySQL clients using a MySQL Cluster as a data source can be
modified to take advantage of the ability to connect with
multiple MySQL servers to achieve load balancing and failover.
For example, Java clients using Connector/J 5.0.6 and later
can use jdbc:mysql:loadbalance:// URLs
(improved in Connector/J 5.1.7) to achieve load balancing
transparently. See Section 21.3, “MySQL Connector/J”, for more
information.
Management clients.
These clients connect to the management server and provide
commands for starting and stopping nodes gracefully,
starting and stopping message tracing (debug versions only),
showing node versions and status, starting and stopping
backups, and so on. Such clients — such as the
ndb_mgm management client supplied with
MySQL Cluster (see
Section 17.4.5, “ndb_mgm — The MySQL Cluster Management Client”) —
are written using the MGM API, a C-language API that
communicates directly with one or more MySQL Cluster
management servers. For more information, see
The MGM API.
Event logs.
MySQL Cluster logs events by category (startup, shutdown,
errors, checkpoints, and so on), priority, and severity. A
complete listing of all reportable events may be found in
Section 17.5.4, “Event Reports Generated in MySQL Cluster”. Event logs are of
two types:
Note
Under normal circumstances, it is necessary and sufficient to
keep and examine only the cluster log. The node logs need be
consulted only for application development and debugging
purposes.
Checkpoint.
Generally speaking, when data is saved to disk, it is said that
a checkpoint has been reached. More specific to Cluster, it is a
point in time where all committed transactions are stored on
disk. With regard to the NDB
storage engine, there are two types of checkpoints which work
together to ensure that a consistent view of the cluster's data
is maintained:
Local Checkpoint (LCP).
This is a checkpoint that is specific to a single node;
however, LCP's take place for all nodes in the cluster more
or less concurrently. An LCP involves saving all of a node's
data to disk, and so usually occurs every few minutes. The
precise interval varies, and depends upon the amount of data
stored by the node, the level of cluster activity, and other
factors.
Global Checkpoint (GCP).
A GCP occurs every few seconds, when transactions for all
nodes are synchronized and the redo-log is flushed to disk.
17.1.2. MySQL Cluster Nodes, Node Groups, Replicas, and Partitions
This section discusses the manner in which MySQL Cluster divides
and duplicates data for storage.
Central to an understanding of this topic are the following
concepts, listed here with brief definitions:
(Data) Node.
An ndbd process, which stores a
replica —that is, a copy of the
partition (see below) assigned to the
node group of which the node is a member.
Each data node should be located on a separate computer. While
it is also possible to host multiple ndbd
processes on a single computer, such a configuration is not
supported.
It is common for the terms “node” and “data
node” to be used interchangeably when referring to an
ndbd process; where mentioned, management
(MGM) nodes (ndb_mgmd processes) and SQL
nodes (mysqld processes) are specified as
such in this discussion.
Node Group.
A node group consists of one or more nodes, and stores
partitions, or sets of replicas (see
next item).
The number of node groups in a MySQL Cluster is not directly
configurable; it is function of the number of data nodes and
of the number of replicas (NumberOfReplicas
configuration parameter), as shown here:
[number_of_node_groups ] = number_of_data_nodes / NumberOfReplicas
Thus, a MySQL Cluster with 4 data nodes has 4 node groups if
NumberOfReplicas is set to 1 in the
config.ini file, 2 node groups if
NumberOfReplicas is set to 2, and 1 node
group if NumberOfReplicas is set to 4.
Replicas are discussed later in this section; for more
information about NumberOfReplicas , see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
Note
All node groups in a MySQL Cluster must have the same number
of data nodes.
Prior to MySQL Cluster NDB 7.0, it was not possible to add new
data nodes to a MySQL Cluster without shutting down the
cluster completely and reloading all of its data. In MySQL
Cluster NDB 7.0 (beginning with MySQL Cluster version NDB
6.4.0), you can add new node groups (and thus new data nodes)
to a running MySQL Cluster — see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”, for
information about how this can be done.
Partition.
This is a portion of the data stored by the cluster. There
are as many cluster partitions as nodes participating in the
cluster. Each node is responsible for keeping at least one
copy of any partitions assigned to it (that is, at least one
replica) available to the cluster.
A replica belongs entirely to a single node; a node can (and
usually does) store several replicas.
MySQL Cluster normally partitions
NDBCLUSTER tables automatically.
However, in MySQL 5.1 and later MySQL Cluster releases, it is
possible to employ user-defined partitioning with
NDBCLUSTER tables. This is
subject to the following limitations:
Only KEY and LINEAR
KEY partitioning schemes can be used with
NDBCLUSTER tables.
The maximum number of partitions that may be definied
explicitly for any NDBCLUSTER
table is 8 per node group. (The number of node groups in a
MySQL Cluster is determined as discussed previously in
this section.)
For more information relating to MySQL Cluster and
user-defined partitioning, see
Section 17.1.5, “Known Limitations of MySQL Cluster”, and
Section 18.5.2, “Partitioning Limitations Relating to Storage Engines”.
Replica.
This is a copy of a cluster partition. Each node in a node
group stores a replica. Also sometimes known as a
partition replica. The number of
replicas is equal to the number of nodes per node group.
The following diagram illustrates a MySQL Cluster with four data
nodes, arranged in two node groups of two nodes each; nodes 1 and
2 belong to node group 0, and nodes 3 and 4 belong to node group
1. Note that only data (ndbd) nodes are shown
here; although a working cluster requires an
ndb_mgm process for cluster management and at
least one SQL node to access the data stored by the cluster, these
have been omitted in the figure for clarity.
The data stored by the cluster is divided into four partitions,
numbered 0, 1, 2, and 3. Each partition is stored — in
multiple copies — on the same node group. Partitions are
stored on alternate node groups:
Partition 0 is stored on node group 0; a primary
replica (primary copy) is stored on node 1, and a
backup replica (backup copy of the
partition) is stored on node 2.
Partition 1 is stored on the other node group (node group 1);
this partition's primary replica is on node 3, and its backup
replica is on node 4.
Partition 2 is stored on node group 0. However, the placing of
its two replicas is reversed from that of Partition 0; for
Partition 2, the primary replica is stored on node 2, and the
backup on node 1.
Partition 3 is stored on node group 1, and the placement of
its two replicas are reversed from those of partition 1. That
is, its primary replica is located on node 4, with the backup
on node 3.
What this means regarding the continued operation of a MySQL
Cluster is this: so long as each node group participating in the
cluster has at least one node operating, the cluster has a
complete copy of all data and remains viable. This is illustrated
in the next diagram.
In this example, where the cluster consists of two node groups of
two nodes each, any combination of at least one node in node group
0 and at least one node in node group 1 is sufficient to keep the
cluster “alive” (indicated by arrows in the diagram).
However, if both nodes from
either node group fail, the remaining two
nodes are not sufficient (shown by the arrows marked out with an
X); in either case, the cluster
has lost an entire partition and so can no longer provide access
to a complete set of all cluster data.
17.1.3. MySQL Cluster Hardware, Software, and Networking Requirements
One of the strengths of MySQL Cluster is that it can be run on
commodity hardware and has no unusual requirements in this regard,
other than for large amounts of RAM, due to the fact that all live
data storage is done in memory. (It is possible to reduce this
requirement using Disk Data tables — see
Section 17.5.10, “MySQL Cluster Disk Data Tables”, for more information
about these.) Naturally, multiple and faster CPUs can enhance
performance. Memory requirements for other Cluster processes are
relatively small.
The software requirements for Cluster are also modest. Host
operating systems do not require any unusual modules, services,
applications, or configuration to support MySQL Cluster. For
supported operating systems, a standard installation should be
sufficient. The MySQL software requirements are simple: all that
is needed is a production release of MySQL
5.1.41-ndb-6.2.19 or 5.1.41-ndb-6.3.32 to have
Cluster support. It is not necessary to compile MySQL yourself
merely to be able to use Cluster. In this
How-To, we assume that you are using the
server binary appropriate to your platform, available via the
MySQL Cluster software downloads page at
http://dev.mysql.com/downloads/select.php?id=14.
For communication between nodes, Cluster supports TCP/IP
networking in any standard topology, and the minimum expected for
each host is a standard 100 Mbps Ethernet card, plus a switch,
hub, or router to provide network connectivity for the cluster as
a whole. We strongly recommend that a MySQL Cluster be run on its
own subnet which is not shared with non-Cluster machines for the
following reasons:
Security.
Communications between Cluster nodes are not encrypted or
shielded in any way. The only means of protecting
transmissions within a MySQL Cluster is to run your Cluster
on a protected network. If you intend to use MySQL Cluster
for Web applications, the cluster should definitely reside
behind your firewall and not in your network's
De-Militarized Zone
(DMZ)
or elsewhere.
See
Section 17.5.9.1, “MySQL Cluster Security and Networking Issues”,
for more information.
Efficiency.
Setting up a MySQL Cluster on a private or protected network
allows the cluster to make exclusive use of bandwidth
between cluster hosts. Using a separate switch for your
MySQL Cluster not only helps protect against unauthorized
access to Cluster data, it also ensures that Cluster nodes
are shielded from interference caused by transmissions
between other computers on the network. For enhanced
reliability, you can use dual switches and dual cards to
remove the network as a single point of failure; many device
drivers support failover for such communication links.
It is also possible to use the high-speed Scalable Coherent
Interface (SCI) with MySQL Cluster, but this is not a requirement.
See Section 17.3.5, “Using High-Speed Interconnects with MySQL Cluster”, for more about
this protocol and its use with MySQL Cluster.
17.1.4. MySQL Cluster Development History
In this section, we discuss changes in the implementation of MySQL
Cluster in MySQL 5.1 and MySQL Cluster NDB 6.x as
compared to MySQL 5.0.
We also discuss our roadmap for further improvements to MySQL
Cluster planned for MySQL Cluster NDB 7.0 and later.
There are a number of significant changes in the implementation of
the NDBCLUSTER storage engine in
mainline MySQL 5.1 releases up to and including MySQL 5.1.23 as
compared to that in MySQL 5.0; MySQL Cluster NDB makes further
changes and improvements in MySQL Cluster in addition to these.
The changes and features most likely to be of interest are shown
in the following table:
17.1.4.1. Development History of MySQL Cluster in MySQL 5.1
A number of features for MySQL Cluster were implemented in MySQL
5.1 through MySQL 5.1.23, when support for MySQL Cluster was
moved to MySQL Cluster NDB. All of the features in the following
list are also available in all MySQL Cluster NDB (6.1 and later)
releases.
Integration of MySQL Cluster into MySQL Replication.
MySQL Cluster Replication makes it possible to replicate
from one MySQL Cluster to another. Updates on any SQL node
(MySQL server) in the cluster acting as the master are
replicated to the slave cluster; the state of the slave
side remains consistent with the cluster acting as the
master. This is sometimes referred to as
asynchronous replication between
clusters, providing geographic
redundancy. It is also possible to replicate
from a MySQL Cluster acting as the master to a standalone
MySQL server acting as the slave, or from a standalone
MySQL master server to to a slave cluster; in either of
these cases, the standalone MySQL server uses a storage
engine other than NDBCLUSTER .
Multi-master replication setups such as circular
replication are also supported.
See Section 17.6, “MySQL Cluster Replication”.
Support for storage of rows on disk.
Storage of NDBCLUSTER table
data on disk is now supported. Indexed columns, including
the primary key hash index, must still be stored in RAM;
however, all other columns can be stored on disk.
See Section 17.5.10, “MySQL Cluster Disk Data Tables”.
Variable-size columns.
In MySQL 5.0, an NDBCLUSTER
table column defined as VARCHAR(255)
used 260 bytes of storage independent of what was stored
in any particular record. In MySQL 5.1 Cluster tables,
only the portion of the column actually taken up by the
record is stored. This makes possible a significant
reduction in space requirements for such columns as
compared to previous release series — by a factor of
up to 5 in many cases.
User-defined partitioning.
Users can define partitions based on columns that are part
of the primary key. It is possible to partition
NDB tables based on
KEY and LINEAR KEY
schemes. This feature is also available for many other
MySQL storage engines, which support additional
partitioning types that are not available with
NDBCLUSTER tables.
For additional general information about user-defined
partitioning in MySQL 5.1, see
Chapter 18, Partitioning. Specifics of partitioning
types are discussed in Section 18.2, “Partition Types”.
The MySQL Server can also determine whether it is possible
to “prune away” some of the partitions from the
WHERE clause, which can greatly speed up
some queries. See Section 18.4, “Partition Pruning”,
for information about designing tables and queries to take
advantage of partition pruning.
Autodiscovery of table schema changes.
In MySQL 5.0, it was necessary to issue a
FLUSH
TABLES statement or a “dummy”
SELECT in order for new
NDBCLUSTER tables or changes
made to schemas of existing
NDBCLUSTER tables on one SQL
node to be visible on the cluster's other SQL nodes.
In MySQL 5.1, this is no longer necessary; new Cluster
tables and changes in the definitions of existing
NDBCLUSTER tables made on one
SQL node are immediately visible to all SQL nodes
connected to the cluster.
Note
When creating a new database, it is still necessary in
MySQL 5.1 to issue a CREATE
DATABASE or
CREATE
SCHEMA statement on each SQL node in the
cluster.
Distribution awareness (NDB API).
Distribution awareness is a
mechanism by which the best data node is automatically
selected to be queried for information. (Conceptually, it
is similar in some ways to partition pruning (see
Section 18.4, “Partition Pruning”). To take advantage
of distribution awareness, you should do the following:
Determine which table column is most likely to be
used for finding matching records.
Make this column part of the table's primary
key.
Explicitly partition the table by
KEY , using this column as the
table' partitioning key.
Following these steps causes records with the same value
for the partitioning column to be stored on the same
partition (that is, in the same node group). When reading
data, transactions are begun on the data node actually
having the desired rows instead of this node being
determined by the usual round-robin mechanism.
Important
In order to see a measureable impact on performance,
the cluster must have at least four data nodes, since,
with only two data nodes, both data nodes have exactly
the same data.
Using distribution awareness can yield performance
increase of as great as 45% when using four data nodes,
and possibly more when using a greater number of data
nodes.
Note
In mainline MySQL 5.1 releases, distribution awareness
was supported only when using the NDB API; support was
added for SQL and API nodes in MySQL Cluster NDB 6.3
(see
Section 17.1.4.4, “MySQL Cluster Development in MySQL Cluster NDB 6.3”,
which includes an example showing how to create a
table in order to take advantage of distribution
awareness).
See Section 17.1.5.11, “Previous MySQL Cluster Issues Resolved in MySQL 5.1, MySQL Cluster NDB
6.x, and MySQL Cluster NDB 7.x”, for
more information.
17.1.4.2. MySQL Cluster Development in MySQL Cluster NDB 6.1
The following list provides an overview of significant feature
additions and changes made in MySQL Cluster NDB 6.1. All of the
changes in this list are also available in MySQL Cluster NDB 6.2
and 6.3 releases. For detailed information about all changes
made in MySQL Cluster NDB 6.1, see
Section 17.7.5, “Changes in MySQL Cluster NDB 6.1”.
Increased number of cluster nodes.
The maximum number of all nodes in a MySQL Cluster has
been increased to 255. For more information, see
Section 17.1.5.10, “Limitations Relating to Multiple MySQL Cluster Nodes”.
Disabling arbitration.
It is now possible to disable arbitration by setting
ArbitrationRank=0 on all cluster
management and SQL nodes. For more information, see
Defining
the Management Server:
ArbitrationRank
and
Defining
SQL and Other API Nodes:
ArbitrationRank .
Additional DUMP commands.
New management client DUMP commands
provide help with tracking transactions, scan
operations, and locks. See
Section 17.5.2, “Commands in the MySQL Cluster Management Client”, and
DUMP Commands, for more
information.
Faster Disk Data backups.
Improvements in backups of Disk Data tables can yield a
10 to 15% increase in backup speed of Disk Data tables.
Batched slave updates.
Batching of updates on cluster replication slaves,
enabled using the
--slave-allow-batching option for
mysqld, increases replication
efficiency. For more information, see
Section 17.6.6, “Starting MySQL Cluster Replication (Single Replication Channel)”.
17.1.4.3. MySQL CLuster Development in MySQL Cluster NDB 6.2
The following list provides an overview of significant feature
additions and changes made in MySQL Cluster NDB 6.2. All of the
changes in this list are also available in MySQL Cluster NDB 6.3
. For more detailed information about all feature changes and
bugfixes made in MySQL Cluster NDB 6.2, see
Section 17.7.4, “Changes in MySQL Cluster NDB 6.2”.
Enhanced backup status reporting.
Backup status reporting has been improved, aided in part
by the introduction of a
BackupReportFrequency configuration
parameter; see
Defining
Data Nodes:
BackupReportFrequency ,
for more information.
Multiple cluster connections per SQL node.
A single MySQL server acting as a MySQL Cluster SQL node
can employ multiple connections to the cluster using the
--ndb-cluster-connection-pool startup
option for mysqld. This option is
described in
MySQL
Cluster-Related Command Options for
mysqld:
--ndb-cluster-connection-pool
option.
New data access interface.
The NdbRecord interface provides a
new and simplified data handler for use in NDB API
applications. See The NdbRecord Interface, for
more information.
New reporting commands.
The new management client REPORT
BackupStatus and REPORT
MemoryUsage commands provide better access to
information about the status of MySQL Cluster backups
and how much memory is being used by MySQL Cluster for
data and index storage. See
Section 17.5.2, “Commands in the MySQL Cluster Management Client”, for
more information about the REPORT
commands. In addition, in-progress status reporting is
provided by the ndb_restore utility;
see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
Improved memory allocation and configuration.
Memory is now allocated by the
NDB kernel to tables on a
page-by-page basis, which significantly reduces the
memory overhead required for maintaining
NDBCLUSTER tables. In
addition, the MaxAllocate
configuration parameter now makes it possible to set the
maximum size of the allocation unit used for table
memory; for more information about this configuration
parameter, see
Defining
Data Nodes:
MaxAllocate .
Choice of fixed-width or variable-width columns.
You can control whether fixed-width or variable-width
storage is used for a given column of an
NDB table by employing of
the COLUMN_FORMAT specifier as part
of the column's definition in a
CREATE TABLE or
ALTER TABLE statement. In
addition, the ability to control whether a given column
of an NDB table is stored
in memory or on disk, using the
STORAGE specifier as part of the
column's definition in a CREATE
TABLE or ALTER
TABLE statement. For more information, see
Section 12.1.17, “CREATE TABLE Syntax”, and
Section 12.1.7, “ALTER TABLE Syntax”.
Controlling management client connections.
The --bind-address cluster management
server startup option makes it possible to restrict
management client connections to
ndb_mgmd to a single host (IP address
or host name) and port, which can make MySQL Cluster
management operations more secure. For more information
about this option, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
Micro-GCPs.
Due to a change in the protocol for handling of global
checkpoints (GCPs handled in this manner sometimes being
referred to as “micro-GCPs”), it is now
possible to control how often the GCI number is updated,
and how often global checkpoints are written to disk,
using the TimeBetweenEpochs
configuration parameter. This improves the reliability
and performance of MySQL Cluster Replication. For more
information, see
Defining
Data Nodes:
TimeBetweenEpochs
and
Defining
Data Nodes:
TimeBetweenEpochsTimeout .
Core online schema change support.
Support for the online ALTER
TABLE operations ADD
COLUMN , ADD INDEX , and
DROP INDEX is available.
When the ONLINE keyword is used, the
ALTER TABLE is
noncopying, which means that indexes do not have to be
re-created, which provides these benefits:
Single user mode is no longer required for
ALTER TABLE
operations that can be performed online.
Transactions can continue during
ALTER TABLE
operations that can be performed online.
Tables being altered online are not locked against
access by other SQL nodes.
However, such tables are locked against other
operations on the same SQL
node for the duration of the
ALTER TABLE . We are
working to overcome this limitation in a future
MySQL Cluster release.
Online CREATE INDEX and
DROP INDEX statements are
also supported. Online changes can be suppressed using
the OFFLINE key word. See
Section 12.1.7, “ALTER TABLE Syntax”,
Section 12.1.13, “CREATE INDEX Syntax”, and
Section 12.1.24, “DROP INDEX Syntax”, for more detailed
information.
mysql.ndb_binlog_index improvements.
More information has been added to the
mysql.ndb_binlog_index table so that
it is possible to determine which originating epochs
have been applied inside an epoch. This is particularly
useful for 3-way replication. See
Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
more information.
Epoch lag control.
The MaxBufferedEpochs data node
configuration parameter provides a means to control the
maximum number of unprocessed epochs by which a
subscribing node can lag. Subscribers which exceed this
number are disconnected and forced to reconnect. For a
discussion of this configuration parameter, see
Defining
Data Nodes:
MaxBufferedEpochs .
Fully automatic database discovery.
It is no longer a requirement for database autodiscovery
that an SQL node already be connected to the cluster at
the time that a database is created on another SQL node,
or for a CREATE DATABASE
or
CREATE
SCHEMA statement to be issued on the new SQL
node after it joins the cluster.
Multiple data node processes per host.
In earlier MySQL Cluster release series, we did not
support MySQL Cluster deployments in production where
more than one ndbd process was run on
a single physical machine. However, beginning with MySQL
Cluster NDB 6.2.0, you can use multiple data node
processes on a single host.
Improved Disk Data filesystem configuration.
As of MySQL Cluster NDB 6.2.17, you can specify default
locations for MySQL Cluster Disk Data data files and
undo log files using the data node configuration
parameters FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . This
eliminates the need to use symbolic links in order to
place Disk Data files separately from other files in
data node filesystems to improve Disk Data performance.
For more information, see
Disk Data filesystem parameters.
Automatic creation of Disk Data log file groups and tablespaces.
Beginning with MySQL Cluster NDB 6.2.17, using the data
node configuration parameters
InitialLogFileGroup and
InitialTablespace , you can cause the
creation of a MySQL Cluster Disk Data log file group,
tablespace, or both, when the cluster is first started.
When using these parameters, no SQL statements are
required to create these Disk Data objects. For more
information, see
Disk
Data object creation parameters.
17.1.4.4. MySQL Cluster Development in MySQL Cluster NDB 6.3
The following list provides an overview of significant feature
additions and changes first made in MySQL Cluster NDB 6.3. For
more detailed information about all feature changes and bugfixes
made in MySQL Cluster NDB 6.3, see
Section 17.7.3, “Changes in MySQL Cluster NDB 6.3”.
Conflict detection and resolution.
It is now possible to detect and resolve conflicts that
arise in multi-master replication scenarios, such as
circular replication, when different masters may try to
update the same row on the slave with different data.
Both “greatest timestamp wins” and
“same timestamp wins” scenarios are
supported. For more information, see
Section 17.6.11, “MySQL Cluster Replication Conflict Resolution”.
Recovery of “one master, many slaves” replication setups.
Recovery of multi-way replication setups (“one
master, many slaves”) is now supported via the
--ndb-log-orig server option and
changes in the mysql.ndb_binlog_index
table. See
Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
more information.
Enhanced selection options for transaction coordinator.
New values and behaviors are introduced for
--ndb_optimized_node_selection allowing
for greater flexibility when an SQL node chooses a
transaction coordinator. For more information, see the
description of
ndb_optimized_node_selection
in Section 17.3.4.3, “MySQL Cluster System Variables”.
Replication heartbeats.
Replication heartbeats facilitate the task of monitoring
and detecting failures in master-slave connections in
real time. This feature is implemented via a new
MASTER_HEARTBEAT_PERIOD =
value clause for
the CHANGE MASTER TO
statement and the addition of two status variables
Slave_heartbeat_period
and
Slave_received_heartbeats .
For more information, see
Section 12.6.2.1, “CHANGE MASTER TO Syntax”.
NDB thread locks.
It is possible to lock NDB
execution threads and maintenance threads (such as file
system and other operating system threads) to specific
CPUs on multiprocessor data node hosts, and to leverage
real-time scheduling.
Improved performance of updates using primary keys or unique keys.
The number of unnecessary reads when performing a
primary key or unique key update has been greatly
reduced. Since it is seldom necessary to read a record
prior to an update, this can yield a considerable
improvement in performance. In addition, primary key
columns are no longer written to when not needed during
update operations.
Batching improvements.
Support of batched DELETE
and UPDATE operations has
been significantly improved. Batching of UPDATE
WHERE... and multiple
DELETE operations is also
now implemented.
Improved SQL statement performance metrics.
The Ndb_execute_count
system status variable measures the number of round
trips made by SQL statements to the
NDB kernel, providing an
improved metric for determining efficiency with which
statements are excuted. For more information, see
MySQL
Cluster Status Variables:
Ndb_execute_count .
Compressed LCPs and backups.
Compressed local checkpoints and backups can save 50% or
more of the disk space used by uncompressed LCPs and
backups. These can be enabled using the two new data
node configuration parameters
CompressedLCP and
CompressedBackup , respectively. See
MySQL
Cluster Status Variables:
CompressedBackup ,
and
MySQL
Cluster Status Variables:
CompressedLCP , for
more information about these parameters.
OPTIMIZE TABLE support with
NDBCLUSTER tables.
OPTIMIZE TABLE is
supported for dynamic columns of in-memory
NDB tables. In such cases,
it is no longer necessary to drop (and possibly to
re-create) a table, or to perform a rolling restart, in
order to recover memory from deleted rows for general
re-use by Cluster. The performance of
OPTIMIZE on Cluster tables can be
tuned by adjusting the value of the
ndb_optimization_delay system
variable, which controls the number of milliseconds to
wait between processing batches of rows by
OPTIMIZE TABLE . In
addition, OPTIMIZE TABLE
on an NDBCLUSTER table can
be interrupted by, for example, killing the SQL thread
performing the OPTIMIZE operation.
Batching of transactions.
It is possible to cause statements occurring within the
same transaction to be run as a batch by setting the
session variable
transaction_allow_batching to
1 or ON . To use
this feature,
autocommit must be set
to 0 or OFF . Batch
sizes can be controlled using the
--ndb-batch-size option for
mysqld. For more information, see
Section 17.3.4.2, “mysqld Command Options for MySQL Cluster”,
and Section 17.3.4.3, “MySQL Cluster System Variables”.
Attribute promotion with ndb_restore.
It is possible using ndb_restore to
restore data reliably from a column of a given type to a
column that uses a “larger” type. This is
sometimes referred to as attribute
promotion. For example, MySQL Cluster backup
data that originated in a
SMALLINT column can be
restored to a MEDIUMINT ,
INT , or
BIGINT column. See
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”,
for more information.
Parallel data node recovery.
Recovery of multiple data nodes can now be done in
parallel, rather than sequentially. In other words,
several data nodes can be restored concurrently, which
can often result in much faster recovery times than when
they are restored one at a time.
Increased local checkpoint efficiency.
Only 2 local checkpoints are stored, rather than 3,
lowering disk space requirements and the size and number
of redo log files.
NDBCLUSTER table persistence control.
Persistence of NDB tables
can be controlled using the session variables
ndb_table_temporary and
ndb_table_no_logging .
ndb_table_no_logging causes
NDB tables not to be
checkpointed to disk;
ndb_table_temporary does the same,
and in addition, no schema files are created. See
Section 17.3.4.1, “MySQL Cluster Server Option and Variable Reference”.
Epoll support (Linux only).
Epoll is an improved method for
handling file descriptors, which is more efficient than
scanning to determine whether a file descriptor has data
to be read. (The term epoll is
specific to Linux and equivalent functionality is known
by other names on other platforms such as Solaris and
FreeBSD.) Currently, MySQL Cluster supports this
functionality on Linux only.
Distribution awareness (SQL nodes).
In MySQL Cluster NDB 6.3, SQL nodes can take advantage
of
distribution
awareness. Here we provide a brief example
showing how to design a table to make a given class of
queries distrubtion-aware. Suppose an
NDBCLUSTER table
t1 has the following schema:
CREATE TABLE t1 (
userid INT NOT NULL,
serviceid INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
data VARCHAR(255)
) ENGINE=NDBCLUSTER;
Suppose further that most of the queries to be used in
our application test values of the
userid column of this table. The form
of such a query looks something like this:
SELECT columns FROM t1
WHERE userid relation value ;
In this query, relation
represents some relational operator, such as
= , < ,
> , and so on. Queries using
IN and a list of values can also be
used:
SELECT columns FROM t1
WHERE userid IN value_list ;
In order to make use of distribution awareness, we need
to make the userid column part of the
table's primary key, then explicitly partition the
table with this column being used as the partitioning
key. (Recall that for a partitioned table having one or
more unique keys, all columns of the table's
partitioning key must also be part of all of the unique
keys — for more information and examples, see
Section 18.5.1, “Partitioning Keys, Primary Keys, and Unique Keys”.)
In other words, the table schema should be equivalent to
the following CREATE
TABLE statement:
CREATE TABLE t1 (
userid INT NOT NULL,
serviceid INT NOT NULL AUTO_INCREMENT,
data VARCHAR(255),
PRIMARY KEY p (userid,serviceid)
) ENGINE=NDBCLUSTER
PARTITION BY KEY(userid);
When the table is partitioned in this way, all rows
having the same userid value are
found on the same node group, and the MySQL Server can
immediately select the optimal node to use as the
transaction coordinator.
Realtime extensions for multiple CPUs.
When running MySQL Cluster data nodes on hosts with
multiple processors, the realtime extensions make it
possible to give priority to the data node process and
control on which CPU cores it should operate. This can
be done using the data node configuration parameters
RealtimeScheduler ,
SchedulerExecutionTimer and
SchedulerSpinTimer . Doing so properly
can significantly lower response times and make them
much more predictable response. For more information
about using these parameters, see
Defining
Data Nodes: Realtime Performance Parameters
Fully automatic database discovery.
It is no longer a requirement for database autodiscovery
that an SQL node already be connected to the cluster at
the time that a database is created on another SQL node,
or for a CREATE DATABASE
or
CREATE
SCHEMA statement to be issued on the new SQL
node after it joins the cluster.
Restoring specific databases, tables, or columns from a MySQL Cluster
backup.
It is now possible to exercise more fine-grained control
when restoring a MySQL Cluster from backup using
ndb_restore. Beginning with MySQL
Cluster NDB 6.3.22, you can choose to restore only
specified tables or databases, or exclude specific
tables or databases from being restored, using the new
ndb_restore options
--include-tables ,
--include-databases ,
--exclude-tables , and
--exclude-databases . Beginning with
MySQL Cluster NDB 6.3.26, it is also possible to restore
to a table having fewer columns than the original using
the --exclude-missing-columns option.
For more information about all of these options, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
Improved Disk Data filesystem configuration.
As of MySQL Cluster NDB 6.3.22, you can specify default
locations for MySQL Cluster Disk Data data files and
undo log files using the data node configuration
parameters FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . This
eliminates the need to use symbolic links in order to
place Disk Data files separately from other files in
data node filesystems to improve Disk Data performance.
For more information, see
Disk Data filesystem parameters.
Automatic creation of Disk Data log file groups and tablespaces.
Beginning with MySQL Cluster NDB 6.3.22, using the data
node configuration parameters
InitialLogFileGroup and
InitialTablespace , you can cause the
creation of a MySQL Cluster Disk Data log file group,
tablespace, or both, when the cluster is first started.
When using these parameters, no SQL statements are
required to create these Disk Data objects. For more
information, see
Disk
Data object creation parameters.
Configuration parameter data dumps.
Starting with MySQL Cluster NDB 6.3.25, the
ndb_config utility supports a
--configinfo option that causes it to
dump a list of all configuration parameters supported by
the cluster, along with brief descriptions, information
about the parameters' default and allowed values,
and the sections of the config.ini
file in which the parameters apply. An additional
--xml switch causes
ndb_config to use XML rather than
plaintext output. Using ndb_config
--configinfo or
ndb_config --configinfo
--xml requires no access to a running
MySQL Cluster, any other programs, or any files. For
more information and examples, see
Section 17.4.6, “ndb_config — Extract MySQL Cluster Configuration Information”.
Per-table reporting of free space on disk.
The
INFORMATION_SCHEMA.FILES
table shows information about used and free space in
MySQL Cluster Disk Data data files, but this information
is not applicable to individual tables. In MySQL Cluster
NDB 6.3.27 and later, the ndb_desc
utility provides two additional columns in its output
that show the amount of space allocated on disk for a
given NDB table as well the
amount of space that remains available for additional
storage of disk-based column data for that table. For
more information, see
Section 17.4.9, “ndb_desc — Describe NDB Tables”.
Improved restart times.
Optimizations in redo log handling and other filesystem
operations introduced in MySQL Cluster NDB 6.3.28 have
the potential to reduce considerably the time required
for restarts. While actual performance benefits observed
in production setups will naturally vary depending on
database size, hardware, and other conditions, our own
preliminary testing has shown that these improvements
can yield startup times that are faster than those
typical of previous MySQL Cluster NDB 6.3 releases by a
factor of 50 or more.
Increased flexibility in online upgrade procedure.
Previously, when performing an upgrade of a running
MySQL cluster, the order in which the types of cluster
nodes had to be upgraded was very strict. However,
beginning with MySQL Cluster NDB 6.3.29, MySQL Cluster
supports online upgrading of API nodes (including MySQL
servers running as SQL nodes) before upgrading
management nodes, data nodes, or both.
New replication conflict resolution strategy.
Beginning with MySQL Cluster NDB 6.3.31, the function
NDB$MAX_DELETE_WIN() is available to
implement “greatest timestamp, delete wins”
conflict resolution. See
NDB$MAX_DELETE_WIN(column_name ),
for more information.
Heartbeat thread policy and priority.
Beginning with MySQL Cluster NDB 6.3.32, a new
configuration parameter
HeartbeatThreadPriority makes it
possible to set the policy and the priority for the
heartbeat thread on management and API nodes.
17.1.4.5. MySQL Cluster Development in MySQL Cluster NDB 7.0
The following list provides an overview of significant feature
additions and changes made in MySQL Cluster NDB 7.0. For more
detailed information about all feature changes and bugfixes made
in MySQL Cluster NDB 7.0, see
Section 17.7.2, “Changes in MySQL Cluster NDB 7.0”.
Important
Early development versions of MySQL Cluster NDB 7.0 were known
as “MySQL Cluster NDB 6.4”, and the first four
releases in this series were identified as MySQL Cluster NDB
6.4.0 through 6.4.3. Any information relating to these MySQL
Cluster NDB 6.4.x releases appearing in this documentation
apply to MySQL Cluster NDB 7.0.
MySQL Cluster NDB 7.0.4 is the fifth MySQL Cluster NDB 7.0
release; it is the successor to MySQL Cluster NDB 6.4.3.
MySQL Cluster on Windows (alpha).
MySQL Cluster is now available on an experimental basis
for Windows operating systems. Features and behavior
comparable to those found on platforms that are already
supported — such as Linux and Solaris — are
planned for MySQL Cluster on Windows. Currently, you must
build from source, although we intend to start making
Windows binaries available in the near future. To enable
MySQL Cluster support on Windows, you must configure the
build using the
WITH_NDBCLUSTER_STORAGE_ENGINE option.
For more information, see
Section 2.5.10, “Installing MySQL from Source on Windows”.
Ability to add nodes and node groups online.
Beginning with MySQL Cluster NDB 6.4.0, it is possible to
add new node groups (and thus new data nodes) to a running
MySQL Cluster without shutting down and reloading the
cluster. As part of enabling this feature, a new command
CREATE NODEGROUP has been added to the
cluster management client and the functionality of the
ALTER ONLINE
TABLE ... REORGANIZE PARTITION SQL statement has
been extended. For more information, see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
Data node multithreading support.
Beginning with MySQL Cluster NDB 6.4.0, a multithreaded
version of the data node daemon, named
ndbmtd, is available for use on data
node hosts with multiple CPU cores. This binary is built
automatically when compiling with MySQL Cluster support;
no additional options other than those needed to provide
MySQL Cluster support are needed when configuring the
build. In most respects, ndbmtd
functions in the same way as ndbd, and
can use the same command-line options and configuration
parameters. In addition, the new
MaxNoOfExecutionThreads configuration
parameter can be used to determine the number of data node
process threads for ndbmtd. For more
information, see
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”.
Note
Disk Data tables are not yet supported for use with
ndbmtd.
Configuration cache.
Formerly, MySQL Cluster configuration was stateless
— that is, configuration information was reloaded
from the cluster's global configuration file (usually
config.ini ) each time
ndb_mgmd was started. Beginning with
MySQL Cluster NDB 6.4.0, the cluster's configuration
is cached internally, and the global configuration file is
no longer automatically re-read when the management server
is restarted. This behavior can be controlled via the
three new management server options
--configdir , --initial ,
and --reload . For more information about
this change, see
Section 17.3.2, “MySQL Cluster Configuration Files”. For more
information about the new management server options, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
Snapshot options for backups.
Beginning with MySQL Cluster NDB 6.4.0, you can determine
when performing a cluster backup whether the backup
matches the state of the data when the backup was started
or when it was completed, using the new options
SNAPSHOTSTART and
SNAPSHOTEND for the management
client's START BACKUP command. See
Section 17.5.3.2, “Using The MySQL Cluster Management Client to Create a Backup”,
for more information.
Dynamic NDB transporter send buffer memory allocation.
Previously, the NDB kernel used a fixed-size send buffer
for every data node in the cluster, which was allocated
when the node started. Because the size of this buffer
could not be changed after the cluster was started, it was
necessary to make it large enough in advance to accomodate
the maximum possible load on any transporter socket.
However, this was an inefficient use of memory, since much
of it often went unused. Beginning with MySQL Cluster NDB
6.4.0, send buffer memory is allocated dynamically from a
memory pool shared between all transporters, which means
that the size of the send buffer can be adjusted as
necessary. This change is reflected by the addition of the
configuration parameters
TotalSendBufferMemory ,
ReservedSendBufferMemory , and
OverLoadLimit , as well as a change in
how the existing SendBufferMemory
configuration parameter is used. For more information, see
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”.
Robust DDL operations.
Beginning with MySQL Cluster NDB 6.4.0, DDL operations
(such as CREATE TABLE or
ALTER TABLE ) are protected
from data node failures; in the event of a data node
failure, such operations are now rolled back gracefully.
Previously, if a data node failed while trying to perform
a DDL operation, the MySQL Cluster data dictionary became
locked and no further DDL statements could be executed
without restarting the cluster.
IPv6 support in MySQL Cluster Replication.
Beginning with MySQL Cluster NDB 6.4.1, IPv6 networking is
supported between MySQL Cluster SQL nodes, which makes it
possible to replicate between instances of MySQL Cluster
using IPv6 addresses. However, IPv6 is supported only for
direct connections between MySQL servers; all connections
within an individual MySQL Cluster
must use IPv4. For more information, see
Section 17.6.3, “Known Issues in MySQL Cluster Replication”.
Restoring specific databases, tables, or columns from a MySQL Cluster
backup.
It is now possible to exercise more fine-grained control
when restoring a MySQL Cluster from backup using
ndb_restore. Beginning with MySQL
Cluster NDB 6.4.3, you can choose to restore only
specified tables or databases, or exclude specific tables
or databases from being restored, using the new
ndb_restore options
--include-tables ,
--include-databases ,
--exclude-tables , and
--exclude-databases . Beginning with MySQL
Cluster NDB 7.0.7, it is also possible to restore to a
table having fewer columns than the original using the
--exclude-missing-columns option. For
more information about all of these options, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
Improved Disk Data filesystem configuration.
As of MySQL Cluster NDB 6.4.3, you can specify default
locations for MySQL Cluster Disk Data data files and undo
log files using the data node configuration parameters
FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . This
eliminates the need to use symbolic links in order to
place Disk Data files separately from other files in data
node filesystems to improve Disk Data performance. For
more information, see
Disk Data filesystem parameters.
Automatic creation of Disk Data log file groups and tablespaces.
Beginning with MySQL Cluster NDB 6.4.3, using the data
node configuration parameters
InitialLogFileGroup and
InitialTablespace , you can cause the
creation of a MySQL Cluster Disk Data log file group,
tablespace, or both, when the cluster is first started.
When using these parameters, no SQL statements are
required to create these Disk Data objects. For more
information, see
Disk
Data object creation parameters.
Improved internal message passing and record handling.
MySQL Cluster NDB 7.0 contains 2 changes that optimize the
use of network connections by addressing the size and
number of messages passed between data nodes, and between
data nodes and API nodes, which can increase MySQL Cluster
and application performance:
Packed reads.
Formerly, each read request signal contained a
list of columns to be retrieved, each of these
column identifiers using 4 bytes within the
message. This meant that the message size
increased as the number of columns being fetched
increased. In addition, in the response from the
data node, each column result was packed to a
4-byte boundary, which resulted in wasted space.
In MySQL Cluster NDB 7.0, messaging for read
operations is optimized in both directions, using
a bitmap in the read request to specify the
columns to be fetched. Where many fields are
requested, this can result in a significant
message size reduction as compared with the old
method. In addition, the 4-byte packing in
responses is no longer used, which means that
smaller fields consume less space.
Long signal transactions.
This enhancement reduces the number of messages
and signals that are sent to data nodes for
complex requests. Prior to MySQL Cluster NDB 7.0,
there was a 100 byte limit on the size of the
request signal, which meant that complex requests
had to be split up between multiple messages prior
to transmission, then reassembled on the receiving
end. In addition to actual payload data, each
message required its own operating system and
protocol overhead such as header information. This
often wasted network bandwidth and data node CPU.
The maximum size of the message is now 32 KB,
which is sufficient to accommodate most queries.
Both of these optimizations are internal to the NDB API, and
so is transparent to applications; this is true whether an
application uses the NDB API directly or does so indirectly
through an SQL node.
Configuration parameter data dumps.
Starting with MySQL Cluster NDB 7.0.6, the
ndb_config utility supports a
--configinfo option that causes it to
dump a list of all configuration parameters supported by
the cluster, along with brief descriptions, information
about the parameters' default and allowed values, and
the sections of the config.ini file
in which the parameters apply. An additional
--xml switch causes
ndb_config to use XML rather than
plaintext output. Using ndb_config
--configinfo or
ndb_config --configinfo
--xml requires no access to a running
MySQL Cluster, any other programs, or any files. For more
information and examples, see
Section 17.4.6, “ndb_config — Extract MySQL Cluster Configuration Information”.
Per-table reporting of free space on disk.
The INFORMATION_SCHEMA.FILES
table shows information about used and free space in MySQL
Cluster Disk Data data files, but this information is not
applicable to individual tables. In MySQL Cluster NDB
7.0.8 and later, the ndb_desc utility
provides two additional columns in its output that show
the amount of space allocated on disk for a given
NDB table as well the amount
of space that remains available for additional storage of
disk-based column data for that table. For more
information, see
Section 17.4.9, “ndb_desc — Describe NDB Tables”.
Improved restart times.
Optimizations in redo log handling and other filesystem
operations introduced in MySQL Cluster NDB 7.0.9 have the
potential to reduce considerably the time required for
restarts. While actual performance benefits observed in
production setups will naturally vary depending on
database size, hardware, and other conditions, our own
preliminary testing has shown that these improvements can
yield startup times that are faster than those typical of
previous MySQL Cluster NDB 7.0 releases by a factor of 50
or more.
--nowait-nodes option for management
servers.
Starting with MySQL Cluster NDB 7.0.10, it is possible to
configure a cluster with two management servers, but to
start the cluster using only one of them by starting the
management node daemon with the
--nowait-nodes option.
The other management server can then be started at a later
time to join the running MySQL Cluster.
Increased flexibility in online upgrade procedure.
Previously, when performing an upgrade of a running MySQL
cluster, the order in which the types of cluster nodes had
to be upgraded was very strict. However, beginning with
MySQL Cluster NDB 7.0.10, MySQL Cluster supports online
upgrading of API nodes (including MySQL servers running as
SQL nodes) online upgrading management nodes, data nodes,
or both.
New replication conflict resolution strategy.
Beginning with MySQL Cluster NDB 7.0.11, the function
NDB$MAX_DELETE_WIN() is available to
implement “greatest timestamp, delete wins”
conflict resolution. See
NDB$MAX_DELETE_WIN(column_name ),
for more information.
Improved lock handling for primary key lookups on
BLOB tables.
A MySQL Cluster table stores all but the first 256 bytes
of any BLOB or
TEXT column values in a
separate BLOB table; when
executing queries against such tables, a shared lock is
obtained. Previously, when the query used a primary key
lookup and took place within a transaction, the lock was
held for the duration of the transaction, even after no
more data was being read from the
NDB table. Now in such cases,
the lock is released when all
BLOB data associated with
the table has been read. (Bug#49190)
Note
A shared lock is also taken for unique key lookups; it is
still the case that this lock is held for the duration of
the transaction.
Heartbeat thread policy and priority.
Beginning with MySQL Cluster NDB 7.0.13, a new
configuration parameter
HeartbeatThreadPriority makes it
possible to set the policy and the priority for the
heartbeat thread on management and API nodes.
17.1.4.6. MySQL Cluster Development in MySQL Cluster NDB 7.1
The following improvements to MySQL Cluster have been made in
MySQL Cluster NDB 7.1.
Important
These features are in early development
phase. Timing, availability, and implementation
details are not guaranteed, and are subject to change at any
time without notice.
Java connectors for MySQL Cluster.
The MySQL Cluster distribution now includes 2 new Java
user APIs, ClusterJ and ClusterJPA. ClusterJ is an
object-relational interface in a manner similar to that of
Java persistence frameworks such as Hibernate. Cluster JPA
is a reimplementation of OpenJPA. ClusterJ uses a backend
library (NdbJTie) that provides access to the
NDB storage engine without
using a MySQL Server connection or JDBC. ClusterJPA also
uses NdbJTie when it improves performance, but can also
process complex queries using JDBC and a MySQL Server
connection, where it can take advantage of the MySQL query
optimizer.
ClusterJ and Cluster JPA can also be made to work with
recent MySQL Cluster NDB 7.0 releases although the necessary
library and JAR files are included only in MySQL Cluster NDB
7.1.1 and later.
MySQL Cluster information database (ndbinfo ).
The ndbinfo information database makes
it possible to obtain real-time characteristics of a MySQL
Cluster by issuing queries from the
mysql client or other MySQL client
applications. ndbinfo provides metadata
specific to MySQL Cluster similarly to how the
INFORMATION_SCHEMA database provides
metadata for the standard MySQL Server. This eliminates
much of the need to read log files, issue
REPORT or DUMP
commands in the ndb_mgm client, or
parse the output of ndb_config in order
to get configuration and status information from a running
MySQL Cluster.
For more information, see
Section 17.5.8, “The ndbinfo MySQL Cluster Information Database”.
Native support for default column values.
Starting with MySQL Cluster NDB 7.1.0, default values for
table columns are stored in the NDB kernel rather than by
the MySQL server as was done previously. This means that
inserts on tables having column value defaults can be
smaller and faster than before, because less data must be
sent from SQL nodes to NDBCLUSTER .
Tables created using previous MySQL Cluster releases can
still be used in MySQL Cluster 7.1.0 and later; however,
they do not support native default values until they are
upgraded. You can upgrade a table with non-native default
values to support native default values using an offline
ALTER TABLE statement.
Further enhancements based on this work that we hope to
implement in future MySQL Cluster releases include the
following:
Decrease NDB space
requirements, since there is no longer any need to
store the default value for every row
Implement REPLACE more
efficiently, because there is no need any longer to
read the current row before replacing it
Provide idempotency in replication for tables having
non-nullable columns
--nowait-nodes option for management
servers.
It is now possible to configure a cluster with two
management servers, but to start the cluster using only
one of them by starting the management node daemon with
the --nowait-nodes
option. The other management server can then be started at
a later time to join the running MySQL Cluster.
Heartbeat thread policy and priority.
Beginning with MySQL Cluster NDB 7.1.2, a new
configuration parameter
HeartbeatThreadPriority makes it
possible to set the policy and the priority for the
heartbeat thread on management and API nodes.
17.1.5. Known Limitations of MySQL Cluster
In the sections that follow, we discuss known limitations in
current releases of MySQL Cluster as compared with the features
available when using the MyISAM and
InnoDB storage engines. If you check the
“Cluster” category in the MySQL bugs database at
http://bugs.mysql.com, you can find known bugs in
the following categories under “MySQL Server:” in the
MySQL bugs database at http://bugs.mysql.com, which
we intend to correct in upcoming releases of MySQL Cluster:
This information is intended to be complete with respect to the
conditions just set forth. You can report any discrepancies that
you encounter to the MySQL bugs database using the instructions
given in Section 1.7, “How to Report Bugs or Problems”. If we do not plan to fix
the problem in MySQL Cluster NDB 6.X or 7.X, we will add it to the
list.
See Section 17.1.5.11, “Previous MySQL Cluster Issues Resolved in MySQL 5.1, MySQL Cluster NDB
6.x, and MySQL Cluster NDB 7.x” for a
list of issues in MySQL Cluster in MySQL 5.0 that
have been resolved in the current version.
17.1.5.1. Noncompliance with SQL Syntax in MySQL Cluster
Some SQL statements relating to certain MySQL features produce
errors when used with NDB tables,
as described in the following list:
Temporary tables.
Temporary tables are not supported. Trying either to
create a temporary table that uses the
NDB storage engine or to
alter an existing temporary table to use
NDB fails with the error
Table storage engine 'ndbcluster' does not
support the create option 'TEMPORARY'.
Indexes and keys in NDB tables.
Keys and indexes on MySQL Cluster tables are subject to
the following limitations:
Column width.
Attempting to create an index on an
NDB table column whose width is
greater than 3072 bytes succeeds, but only the first
3072 bytes are actually used for the index. In such
cases, a warning Specified key was too
long; max key length is 3072 bytes is
issued, and a SHOW CREATE
TABLE statement shows the length of the
index as 3072.
TEXT and
BLOB columns.
You cannot create indexes on
NDB table columns that
use any of the TEXT or
BLOB data types.
FULLTEXT indexes.
The NDB storage engine
does not support FULLTEXT indexes,
which are possible for MyISAM
tables only.
However, you can create indexes on
VARCHAR columns of
NDB tables.
Prefixes.
There are no prefix indexes; only entire columns can
be indexed. (The size of an NDB
column index is always the same as the width of the
column in bytes, up to and including 3072 bytes, as
described earlier in this section. Also see
Section 17.1.5.6, “Unsupported or Missing Features in MySQL Cluster”,
for additional information.)
BIT columns.
A BIT column cannot be
a primary key, unique key, or index, nor can it be
part of a composite primary key, unique key, or index.
AUTO_INCREMENT columns.
Like other MySQL storage engines, the
NDB storage engine can
handle a maximum of one
AUTO_INCREMENT column per table.
However, in the case of a Cluster table with no
explicit primary key, an
AUTO_INCREMENT column is
automatically defined and used as a
“hidden” primary key. For this reason,
you cannot define a table that has an explicit
AUTO_INCREMENT column unless that
column is also declared using the PRIMARY
KEY option. Attempting to create a table
with an AUTO_INCREMENT column that
is not the table's primary key, and using the
NDB storage engine, fails
with an error.
MySQL Cluster and geometry data types.
Geometry datatypes (WKT and
WKB ) are supported in
NDB tables in MySQL 5.1
(including MySQL Cluster NDB 6.X and 7.X through 7.1).
However, spatial indexes are not supported.
Character sets and binary log files.
Currently, the ndb_apply_status and
ndb_binlog_index tables are created
using the latin1 (ASCII) character set.
Because names of binary logs are recorded in this table,
binary log files named using non-Latin characters are not
referenced correctly in these tables. This is a known
issue, which we are working to fix. (Bug#50226)
To work around this problem, use only Latin-1 characters
when naming binary log files or setting any the
--basedir ,
--log-bin , or
--log-bin-index options.
Creating NDBCLUSTER tables with
user-defined partitioning.
Support for user-defined partitioning for MySQL Cluster in
MySQL 5.1 (including MySQL Cluster NDB 6.X and 7.X through
7.1) is restricted to [LINEAR ]
KEY partitioning. Beginning with MySQL
5.1.12, using any other partitioning type with
ENGINE=NDB or
ENGINE=NDBCLUSTER in a
CREATE TABLE statement
results in an error.
Default partitioning scheme.
As of MySQL 5.1.6, all MySQL Cluster tables are by default
partitioned by KEY using the table's
primary key as the partitioning key. If no primary key is
explicitly set for the table, the “hidden”
primary key automatically created by the
NDBCLUSTER storage engine is
used instead. For additional discussion of these and
related issues, see Section 18.2.4, “KEY Partitioning”.
Beginning with MySQL Cluster NDB 6.2.18, MySQL Cluster NDB
6.3.25, and MySQL Cluster NDB 7.0.6,
CREATE TABLE and
ALTER TABLE statements that
would cause a user-partitioned
NDBCLUSTER table not to meet
either or both of the following two requirements are
disallowed, and fail with an error (Bug#40709):
The table must have an explicit primary key.
All columns listed in the table's partitioning
expression must be part of the primary key.
Exception.
If a user-partitioned
NDBCLUSTER table is created
using an empty column-list (that is, using
PARTITION BY [LINEAR] KEY() ), then no
explicit primary key is required.
Maximum number of partitions for NDBCLUSTER
tables.
The maximum number of partitions that can defined for a
NDBCLUSTER table when
employing user-defined partitioning is 8 per node group.
(See Section 17.1.2, “MySQL Cluster Nodes, Node Groups, Replicas, and Partitions”, for
more information about MySQL Cluster node groups.
DROP PARTITION not supported.
It is not possible to drop partitions from
NDB tables using
ALTER TABLE ... DROP PARTITION . The
other partitioning extensions to
ALTER TABLE —
ADD PARTITION , REORGANIZE
PARTITION , and COALESCE
PARTITION — are supported for Cluster
tables, but use copying and so are not optimised. See
Section 18.3.1, “Management of RANGE and LIST
Partitions” and
Section 12.1.7, “ALTER TABLE Syntax”.
Row-based replication.
When using row-based replication with MySQL Cluster,
binary logging cannot be disabled. That is, the
NDB storage engine ignores
the value of sql_log_bin .
(Bug#16680)
17.1.5.2. Limits and Differences of MySQL Cluster from Standard MySQL Limits
In this section, we list limits found in MySQL Cluster that
either differ from limits found in, or that are not found in,
standard MySQL.
Memory usage and recovery.
Memory consumed when data is inserted into an
NDB table is not automatically
recovered when deleted, as it is with other storage engines.
Instead, the following rules hold true:
A DELETE statement on an
NDB table makes the memory
formerly used by the deleted rows available for re-use by
inserts on the same table only. This memory cannot be used
by other NDB tables.
A DROP TABLE or
TRUNCATE TABLE operation on
an NDB table frees the memory
that was used by this table for re-use by any
NDB table, either by the same
table or by another NDB table.
Memory freed by DELETE
operations but still allocated to a specific table can also
be made available for general re-use by performing a rolling
restart of the cluster. See
Section 17.2.6.1, “Performing a Rolling Restart of a MySQL Cluster”.
Beginning with MySQL Cluster NDB 6.3.7, this limitation can
be overcome using OPTIMIZE
TABLE . See
Section 17.1.5.11, “Previous MySQL Cluster Issues Resolved in MySQL 5.1, MySQL Cluster NDB
6.x, and MySQL Cluster NDB 7.x”, for
more information.
Limits imposed by the cluster's configuration.
A number of hard limits exist which are configurable, but
available main memory in the cluster sets limits. See the
complete list of configuration parameters in
Section 17.3.2, “MySQL Cluster Configuration Files”. Most
configuration parameters can be upgraded online. These
hard limits include:
Database memory size and index memory size
(DataMemory and
IndexMemory , respectively).
DataMemory is allocated as 32KB
pages. As each DataMemory page is
used, it is assigned to a specific table; once
allocated, this memory cannot be freed except by
dropping the table.
See Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for
further information about DataMemory
and IndexMemory .
The maximum number of operations that can be performed
per transaction is set using the configuration
parameters
MaxNoOfConcurrentOperations and
MaxNoOfLocalOperations .
Note
Bulk loading, TRUNCATE
TABLE , and ALTER
TABLE are handled as special cases by
running multiple transactions, and so are not subject
to this limitation.
Different limits related to tables and indexes. For
example, the maximum number of ordered indexes in the
cluster is determined by
MaxNoOfOrderedIndexes , and the
maximum number of ordered inexes per table is 16.
Node and data object maximums.
The following limits apply to numbers of cluster nodes and
metadata objects:
The maximum number of data nodes is 48.
A data node must have a node ID in the range of 1 to 48,
inclusive. (Previous to MySQL Cluster NDB 6.1.1,
management and API nodes were restricted to the range 1
to 63 inclusive as a node ID; starting with MySQL
Cluster NDB 6.1.1, management and API nodes may use node
IDs in the range 1 to 255, inclusive.)
Prior to MySQL Cluster NDB 6.1.1, the total maximum
number of nodes in a MySQL Cluster was 63. Beginning
with MySQL Cluster NDB 6.1.1, the total maximum number
of nodes in a MySQL Cluster is 255. In either case, this
number includes all SQL nodes (MySQL Servers), API nodes
(applications accessing the cluster other than MySQL
servers), data nodes, and management servers.
The maximum number of metadata objects in current
versions of MySQL Cluster is 20320. This limit is
hard-coded.
See Section 17.1.5.11, “Previous MySQL Cluster Issues Resolved in MySQL 5.1, MySQL Cluster NDB
6.x, and MySQL Cluster NDB 7.x”,
for more information.
17.1.5.3. Limits Relating to Transaction Handling in MySQL Cluster
A number of limitations exist in MySQL Cluster with regard to
the handling of transactions. These include the following:
Transaction isolation level.
The NDBCLUSTER storage engine
supports only the READ
COMMITTED transaction isolation level.
(InnoDB , for example, supports
READ COMMITTED ,
READ UNCOMMITTED ,
REPEATABLE READ , and
SERIALIZABLE .) See
Section 17.5.3.4, “MySQL Cluster Backup Troubleshooting”,
for information on how this can affect backing up and
restoring Cluster databases.)
Transactions and BLOB or
TEXT columns.
NDBCLUSTER stores only part
of a column value that uses any of MySQL's
BLOB or
TEXT data types in the
table visible to MySQL; the remainder of the
BLOB or
TEXT is stored in a
separate internal table that is not accessible to MySQL.
This gives rise to two related issues of which you should
be aware whenever executing
SELECT statements on tables
that contain columns of these types:
For any SELECT from a
MySQL Cluster table: If the
SELECT includes a
BLOB or
TEXT column, the
READ COMMITTED
transaction isolation level is converted to a read with
read lock. This is done to guarantee consistency.
Prior to MySQL Cluster NDB 7.0.12, for any
SELECT which used a
primary key lookup or unique key lookup to retrieve any
columns that used any of the
BLOB or
TEXT data types and that
was executed within a transaction, a shared read lock
was held on the table for the duration of the
transaction — that is, until the transaction was
either committed or aborted.
In MySQL Cluster NDB 7.0.12 and later, for primary key
lookups, the lock is released as soon as all
BLOB or
TEXT data has been read.
(Bug#49190) However, for unique key lookups, the shared
lock continues to be held for the lifetime of the
transaction.
This issue does not occur for queries that use index or
table scans, even against
NDB tables having
BLOB or
TEXT columns.
For example, consider the table t
defined by the following CREATE
TABLE statement:
CREATE TABLE t (
a INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
b INT NOT NULL,
c INT NOT NULL,
d TEXT,
INDEX i(b),
UNIQUE KEY u(c)
) ENGINE = NDB,
Either of the following queries on t
causes a shared read lock, because the first query uses
a primary key lookup and the second uses a unique key
lookup:
SELECT * FROM t WHERE a = 1;
SELECT * FROM t WHERE c = 1;
However, none of the four queries shown here causes a
shared read lock:
SELECT * FROM t WHERE b 1;
SELECT * FROM t WHERE d = '1';
SELECT * FROM t;
SELECT b,c WHERE a = 1;
This is because, of these four queries, the first uses
an index scan, the second and third use table scans, and
the fourth, while using a primary key lookup, does not
retrieve the value of any
BLOB or
TEXT columns.
You can help minimize issues with shared read locks by
avoiding queries that use unique key lookups (or primary
key lookups in MySQL Cluster NDB 7.0.11 and earlier)
that retrieve BLOB or
TEXT columns, or, in
cases where such queries are not avoidable, by
committing transactions as soon as possible afterwards.
Transactions and memory usage.
As noted elsewhere in this chapter, MySQL Cluster does not
handle large transactions well; it is better to perform a
number of small transactions with a few operations each
than to attempt a single large transaction containing a
great many operations. Among other considerations, large
transactions require very large amounts of memory. Because
of this, the transactional behavior of a number of MySQL
statements is effected as described in the following list:
TRUNCATE TABLE is not
transactional when used on
NDB tables. If a
TRUNCATE TABLE fails to
empty the table, then it must be re-run until it is
successful.
DELETE FROM (even with no
WHERE clause) is
transactional. For tables containing a great many rows,
you may find that performance is improved by using
several DELETE FROM ... LIMIT ...
statements to “chunk” the delete operation.
If your objective is to empty the table, then you may
wish to use TRUNCATE
TABLE instead.
LOAD DATA statements.
LOAD DATA
INFILE is not transactional when used on
NDB tables.
Important
When executing a
LOAD DATA
INFILE statement, the
NDB engine performs
commits at irregular intervals that enable better
utilization of the communication network. It is not
possible to know ahead of time when such commits take
place.
LOAD DATA FROM MASTER is not
supported in MySQL Cluster.
ALTER TABLE and transactions.
When copying an NDB table
as part of an ALTER
TABLE , the creation of the copy is
nontransactional. (In any case, this operation is
rolled back when the copy is deleted.)
Transactions and the COUNT() function.
When using MySQL Cluster Replication, it is not possible
to guarantee the transactional consistency of the
COUNT() function on the slave. In other
words, when performing on the master a series of
statements (INSERT ,
DELETE , or both) that
changes the number of rows in a table within a single
transaction, executing SELECT COUNT(*) FROM
table queries on the
slave may yield intermediate results. This is due to the
fact that SELECT COUNT(...) may perform
dirty reads, and is not a bug in the
NDB storage engine. (See
Bug#31321 for more information.)
17.1.5.4. MySQL Cluster Error Handling
Starting, stopping, or restarting a node may give rise to
temporary errors causing some transactions to fail. These
include the following cases:
Temporary errors.
When first starting a node, it is possible that you may
see Error 1204 Temporary failure, distribution
changed and similar temporary errors.
Errors due to node failure.
The stopping or failure of any data node can result in a
number of different node failure errors. (However, there
should be no aborted transactions when performing a
planned shutdown of the cluster.)
In either of these cases, any errors that are generated must be
handled within the application. This should be done by retrying
the transaction.
See also Section 17.1.5.2, “Limits and Differences of MySQL Cluster from Standard MySQL Limits”.
17.1.5.5. Limits Associated with Database Objects in MySQL Cluster
Some database objects such as tables and indexes have different
limitations when using the
NDBCLUSTER storage engine:
Identifiers.
Database names, table names and attribute names cannot be
as long in NDB tables as when
using other table handlers. Attribute names are truncated
to 31 characters, and if not unique after truncation give
rise to errors. Database names and table names can total a
maximum of 122 characters. In other words, the maximum
length for an NDB table name
is 122 characters, less the number of characters in the
name of the database of which that table is a part.
Table names containing special characters.
NDB tables whose names
contain characters other than letters, numbers, dashes,
and underscores and which are created on one SQL node were
not always discovered correctly by other SQL nodes.
(Bug#31470)
Note
This issue was fixed in MySQL 5.1.23, MySQL Cluster NDB
6.2.7, and MySQL Cluster NDB 6.3.4.
Number of database objects.
The maximum number of all
NDB database objects in a
single MySQL Cluster — including databases, tables,
and indexes — is limited to 20320.
Attributes per table.
The maximum number of attributes (that is, columns and
indexes) per table is limited to 128.
Attributes per key.
The maximum number of attributes per key is 32.
Row size.
The maximum permitted size of any one row is 8KB. Note
that each BLOB or
TEXT column contributes 256
+ 8 = 264 bytes towards this total.
17.1.5.6. Unsupported or Missing Features in MySQL Cluster
A number of features supported by other storage engines are not
supported for NDB tables. Trying to
use any of these features in MySQL Cluster does not cause errors
in or of itself; however, errors may occur in applications that
expects the features to be supported or enforced:
Foreign key constraints.
The foreign key construct is ignored, just as it is in
MyISAM tables.
Index prefixes.
Prefixes on indexes are not supported for
NDBCLUSTER tables. If a prefix is used
as part of an index specification in a statement such as
CREATE TABLE ,
ALTER TABLE , or
CREATE INDEX , the prefix is
ignored.
OPTIMIZE operations.
OPTIMIZE operations are not supported.
Beginning with MySQL Cluster NDB 6.3.7, this limitation has
been lifted. See
Section 17.1.5.11, “Previous MySQL Cluster Issues Resolved in MySQL 5.1, MySQL Cluster NDB
6.x, and MySQL Cluster NDB 7.x”, for
more information.
LOAD TABLE ... FROM MASTER .
LOAD TABLE ... FROM MASTER is not
supported.
Savepoints and rollbacks.
Savepoints and rollbacks to savepoints are ignored as in
MyISAM .
Durability of commits.
There are no durable commits on disk. Commits are
replicated, but there is no guarantee that logs are
flushed to disk on commit.
Replication.
Statement-based replication is not supported. Use
--binlog-format=ROW (or
--binlog-format=MIXED ) when
setting up cluster replication. See
Section 17.6, “MySQL Cluster Replication”, for more
information.
InnoDB plugin not supported.
Currently, MySQL Cluster is not compatible with the
InnoDB Plugin . You must use the version
of InnoDB that is supplied
with the MySQL Server. See
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”, for
information about enabling
InnoDB storage engine support
with MySQL Cluster.
17.1.5.7. Limitations Relating to Performance in MySQL Cluster
The following performance issues are specific to or especially
pronounced in MySQL Cluster:
Range scans.
There are query performance issues due to sequential
access to the NDB storage
engine; it is also relatively more expensive to do many
range scans than it is with either
MyISAM or InnoDB .
Reliability of Records in range .
The Records in range statistic is
available but is not completely tested or officially
supported. This may result in nonoptimal query plans in
some cases. If necessary, you can employ USE
INDEX or FORCE INDEX to alter
the execution plan. See Section 12.2.8.2, “Index Hint Syntax”, for
more information on how to do this.
Unique hash indexes.
Unique hash indexes created with USING
HASH cannot be used for accessing a table if
NULL is given as part of the key.
17.1.5.8. Issues Exclusive to MySQL Cluster
The following are limitations specific to the
NDBCLUSTER storage engine:
Machine architecture.
All machines used in the cluster must have the same
architecture. That is, all machines hosting nodes must be
either big-endian or little-endian, and you cannot use a
mixture of both. For example, you cannot have a management
node running on a PowerPC which directs a data node that
is running on an x86 machine. This restriction does not
apply to machines simply running mysql
or other clients that may be accessing the cluster's SQL
nodes.
Binary logging.
MySQL Cluster has the following limitations or
restrictions with regard to binary logging:
sql_log_bin has no
effect on data operations; however, it is supported for
schema operations.
MySQL Cluster cannot produce a binlog for tables having
BLOB columns but no
primary key.
Only the following schema operations are logged in a
cluster binlog which is not on the
mysqld executing the statement:
See also
Section 17.1.5.10, “Limitations Relating to Multiple MySQL Cluster Nodes”.
17.1.5.9. Limitations Relating to MySQL Cluster Disk Data StorageDisk Data object maxmimums and minimums.
Disk data objects are subject to the following maximums and
minimums:
Maximum number of tablespaces:
232 (4294967296)
Maximum number of data files per tablespace:
216 (65535)
The theoretical maximum number of extents per tablespace
data file is 216 (65525);
however, for practical purposes, the recommended maximum
number of extents per data file is
28 (32768).
Maximum data file size: The theoretical limit is 64G;
however, in MySQL 5.1 (including MySQL Cluster NDB 6.X and
7.X through 7.1), the practical upper limit is 32G. This is
equivalent to 32768 extents of 1M each.
The minimum and maximum possible sizes of extents for
tablespace data files are 32K and 2G, respectively. See
Section 12.1.18, “CREATE TABLESPACE Syntax”, for more information.
Disk Data tables and diskless mode.
Use of Disk Data tables is not supported when running the
cluster in diskless mode. Beginning with MySQL 5.1.12, it is
disallowed altogether. (Bug#20008)
17.1.5.10. Limitations Relating to Multiple MySQL Cluster NodesMultiple SQL nodes.
The following are issues relating to the use of multiple MySQL
servers as MySQL Cluster SQL nodes, and are specific to the
NDBCLUSTER storage engine:
No distributed table locks.
A LOCK TABLES works only
for the SQL node on which the lock is issued; no other SQL
node in the cluster “sees” this lock. This is
also true for a lock issued by any statement that locks
tables as part of its operations. (See next item for an
example.)
ALTER TABLE operations.
ALTER TABLE is not fully
locking when running multiple MySQL servers (SQL nodes).
(As discussed in the previous item, MySQL Cluster does not
support distributed table locks.)
Multiple management nodes.
When using multiple management servers:
You must give nodes explicit IDs in connectstrings because
automatic allocation of node IDs does not work across
multiple management servers.
You must take extreme care to have the same configurations
for all management servers. No special checks for this are
performed by the cluster.
Multiple network addresses.
Multiple network addresses per data node are not supported.
Use of these is liable to cause problems: In the event of a
data node failure, an SQL node waits for confirmation that the
data node went down but never receives it because another
route to that data node remains open. This can effectively
make the cluster inoperable.
Note
It is possible to use multiple network hardware
interfaces (such as Ethernet cards) for a
single data node, but these must be bound to the same address.
This also means that it not possible to use more than one
[tcp] section per connection in the
config.ini file. See
Section 17.3.2.8, “MySQL Cluster TCP/IP Connections”, for more
information.
17.1.5.11. Previous MySQL Cluster Issues Resolved in MySQL 5.1, MySQL Cluster NDB
6.x, and MySQL Cluster NDB 7.x
A number of limitations and related issues existing in earlier
versions of MySQL Cluster have been resolved:
Variable-length column support.
The NDBCLUSTER storage engine
now supports variable-length column types for in-memory
tables.
Previously, for example, any Cluster table having one or
more VARCHAR fields which
contained only relatively small values, much more memory and
disk space were required when using the
NDBCLUSTER storage engine than
would have been the case for the same table and data using
the MyISAM engine. In other words, in the
case of a VARCHAR column,
such a column required the same amount of storage as a
CHAR column of the same size.
In MySQL 5.1, this is no longer the case for in-memory
tables, where storage requirements for variable-length
column types such as VARCHAR
and BINARY are comparable to those for
these column types when used in MyISAM
tables (see Section 10.5, “Data Type Storage Requirements”).
Replication with MySQL Cluster.
It is now possible to use MySQL replication with Cluster
databases. For details, see
Section 17.6, “MySQL Cluster Replication”.
Circular Replication.
Circular replication is also supported with MySQL Cluster,
beginning with MySQL 5.1.18. See
Section 17.6.10, “MySQL Cluster Replication — Multi-Master and Circular Replication”.
auto_increment_increment and
auto_increment_offset .
The
auto_increment_increment
and auto_increment_offset
server system variables are supported for Cluster
replication beginning with MySQL 5.1.20, MySQL Cluster NDB
6.2.5, and MySQL Cluster 6.3.2.
Database autodiscovery and online schema changes.
Autodiscovery of databases is now supported for multiple
MySQL servers accessing the same MySQL Cluster. Formerly,
autodiscovery in MySQL Cluster 5.1 and MySQL Cluster NDB
6.x releases required that a given
mysqld was already running and
connected to the cluster at the time that the database was
created on a different mysqld —
in other words, when a mysqld process
connected to the cluster after a database named
db_name was created, it was
necessary to issue a CREATE DATABASE
db_name or
CREATE SCHEMA
db_name statement on
the “new” MySQL server when it first
accesseed that MySQL Cluster. Beginning with MySQL Cluster
NDB 6.2.16 and MySQL Cluster NDB 6.3.18, such a
CREATE statement is no longer required.
(Bug#39612)
This also means that online schema changes in
NDB tables are now possible.
That is, the result of operations such as
ALTER TABLE and
CREATE INDEX performed on one
SQL node in the cluster are now visible to the cluster's
other SQL nodes without any additional action being taken.
Backup and restore between architectures.
Beginning with MySQL 5.1.10, it is possible to perform a
Cluster backup and restore between different
architectures. Previously — for example — you
could not back up a cluster running on a big-endian
platform and then restore from that backup to a cluster
running on a little-endian system. (Bug#19255)
Character set directory.
Beginning with MySQL 5.1.10, it is possible to install
MySQL with Cluster support to a nondefault location and
change the search path for font description files using
either the --basedir or
--character-sets-dir
options. (Previously, ndbd in MySQL 5.1
searched only the default path — typically
/usr/local/mysql/share/mysql/charsets
— for character sets.)
Multiple management servers.
In MySQL 5.1 (including all MySQL Cluster NDB 6.x
versions), it is no longer necessary, when running
multiple management servers, to restart all the
cluster's data nodes to enable the management nodes
to see one another.
Also, when using multiple management servers and starting
concurrently several API nodes (possibly including one or
more SQL nodes) whose connectstrings listed the management
servers in different order, it was possible for 2 API nodes
to be assigned the same node ID. This issue is resolved in
MySQL Cluster NDB 6.2.17, 6.3.23, and 6.4.3. (Bug#42973)
Multiple data node processes per host.
Beginning with MySQL Cluster NDB 6.2.0, you can use
multiple data node processes on a single host. (In MySQL
Cluster NDB 6.1, MySQL 5.1, and earlier release series, we
did not support production MySQL Cluster deployments in
which more than one ndbd process was
run on a single physical machine.)
In addition, MySQL Cluster NDB 7.0 introduces support for
multi-threaded data nodes (ndbmtd). See
Section 17.1.4.5, “MySQL Cluster Development in MySQL Cluster NDB 7.0”, and
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”, for more
information.
Length of CREATE TABLE statements.
CREATE TABLE statements may
be no more than 4096 characters in length. This
limitation affects MySQL 5.1.6, 5.1.7, and 5.1.8
only. (See Bug#17813)
IGNORE and REPLACE
functionality.
In MySQL 5.1.7 and earlier,
INSERT
IGNORE ,
UPDATE
IGNORE , and
REPLACE were supported only
for primary keys, but not for unique keys. It was possible
to work around this issue by removing the constraint, then
dropping the unique index, performing any inserts, and
then adding the unique index again.
This limitation was removed for
INSERT
IGNORE and REPLACE
in MySQL 5.1.8. (See Bug#17431.)
AUTO_INCREMENT columns.
In MySQL 5.1.10 and earlier versions, the maximum number
of tables having AUTO_INCREMENT columns
— including those belonging to hidden primary keys
— was 2048.
This limitation was lifted in MySQL 5.1.11.
Maximum number of cluster nodes.
Prior to MySQL Cluster NDB 6.1.1, the total maximum number
of nodes in a MySQL Cluster was 63, including all SQL
nodes (MySQL Servers), API nodes (applications accessing
the cluster other than MySQL servers), data nodes, and
management servers.
Starting with MySQL Cluster NDB 6.1.1, the total maximum
number of nodes in a MySQL Cluster is 255, including all SQL
nodes (MySQL Servers), API nodes (applications accessing the
cluster other than MySQL servers), data nodes, and
management servers. The total number of data nodes and
management nodes beginning with this version is 63, of which
up to 48 can be data nodes.
Note
The limitation that a data node cannot have a node ID
greater than 49 continues to apply.
Recovery of memory from deleted rows.
Beginning with MySQL Cluster NDB 6.3.7, memory can be
reclaimed from an NDB table
for reuse with any NDB table
by employing OPTIMIZE
TABLE , subject to the following limitations:
You can regulate the effects of OPTIMIZE
on performance by adjusting the value of the global system
variable ndb_optimization_delay , which
sets the number of milliseconds to wait between batches of
rows being processed by OPTIMIZE . The
default value is 10 milliseconds. It is possible to set a
lower value (to a minimum of 0 ), but not
recommended. The maximum is 100000 milliseconds (that is,
100 seconds).
Implicit Rollbacks.
Prior to MySQL Cluster NDB 6.2.17 and MySQL Cluster NDB
6.3.19, MySQL Cluster did not automtically roll back a
transaction that was aborted by a duplicate key or similar
error, and subsequent statements raised ERROR
1296 (HY000): Got error 4350 'Transaction already aborted'
from NDBCLUSTER. In such cases, it was
necessary to issue an explicit
ROLLBACK
statement first, and then to retry the entire transaction.
Beginning with MySQL Cluster NDB 6.2.17 and MySQL Cluster
NDB 6.3.19, this limitation has been removed; now, an error
which causes a transaction to be aborted generates an
implicit rollback of the entire transaction. This is logged
with the warning Storage engine NDB does not
support rollback for this statement. Transaction rolled back
and must be restarted. A statement subsequent to
this starts a new transaction. (Bug#32656)
Note
The NDBCLUSTER storage engine
does not support partial transactions or partial rollbacks
of transactions in any version of MySQL Cluster.
Number of tables.
Previously, the maximum number of
NDBCLUSTER tables in a single
MySQL Cluster was 1792, but this is no longer the case in
MySQL 5.1 and later MySQL Cluster releases. However, the
number of tables is still included in the total maximum
number of NDBCLUSTER database
objects (20320). (See
Section 17.1.5.5, “Limits Associated with Database Objects in MySQL Cluster”.)
DDL operations.
Beginning with MySQL Cluster NDB 6.4.0, DDL operations
(such as CREATE TABLE or
ALTER TABLE ) are protected
from data node failures. Previously, if a data node failed
while trying to perform one of these, the data dictionary
became locked and no further DDL statements could be
executed without restarting the cluster (Bug#36718).
Adding and dropping of data nodes.
In MySQL Cluster NDB 6.3 and previous versions of MySQL
Cluster, the online adding or dropping of data nodes was
not possible; such operations required a complete shutdown
and restart of the entire cluster. In MySQL Cluster NDB
7.0 (beginning with MySQL Cluster NDB 6.4.0) and later
MySQL Cluster release series, it is possible to add new
data nodes to a running MySQL Cluster by performing a
rolling restart, so that the cluster and the data stored
in it remain available to applications.
When planning to increase the number of data nodes in the
cluster online in MySQL Cluster NDB 7.0 or MySQL Cluster NDB
7.1, you should be aware of and take into account the
following issues:
New data nodes can be added online to a MySQL Cluster
only as part of a new node group.
New data nodes can be added online, but cannot yet be
dropped online. Reducing the number of data nodes still
requires a system restart of the cluster.
As in previous MySQL Cluster releases, it is not
possible to change online either the number of replicas
(NoOfReplicas configuration
parameter) or the number of data nodes per node group.
These changes require a system restart.
Redistribution of existing cluster data using the new
data nodes is not automatic; however, this can be
accomplished using simple SQL statements in the
mysql client or other MySQL client
application once the nodes have been added. During this
procedure, it is not possible to perform DDL operations,
although DML operations can continue as normal.
The distribution of new cluster data (that is, data
stored in the cluster after the new
nodes have been added) uses the new nodes without manual
intervention.
For more information, see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
Native support for default column values.
Starting with MySQL Cluster NDB 7.1.0, default values for
table columns are stored by
NDBCLUSTER , rather than by
the MySQL server as was previously the case. Because less
data must be sent from an SQL node to the data nodes,
inserts on tables having column value defaults can be
performed more efficiently than before.
Tables created using previous MySQL Cluster releases can
still be used in MySQL Cluster 7.1.0 and later, although
they do not support native default values and continue to
use defaults supplied by the MySQL server until they are
upgraded. This can be done by means of an offline
ALTER TABLE statement.
Important
You cannot set or change a table column's default
value using an online ALTER
TABLE operation
17.2. MySQL Cluster Multi-Computer How-To
This section is a “How-To” that describes the basics
for how to plan, install, configure, and run a MySQL Cluster.
Whereas the examples in
Section 17.3, “MySQL Cluster Configuration” provide more in-depth
information on a variety of clustering options and configuration,
the result of following the guidelines and procedures outlined here
should be a usable MySQL Cluster which meets the
minimum requirements for availability and
safeguarding of data.
This section covers hardware and software requirements; networking
issues; installation of MySQL Cluster; configuration issues;
starting, stopping, and restarting the cluster; loading of a sample
database; and performing queries.
Basic assumptions.
This How-To makes the following
assumptions:
The cluster is to be set up with four nodes, each on a separate
host, and each with a fixed network address on a typical
Ethernet network as shown here:
This may be made clearer in the following diagram:
In the interest of simplicity (and reliability), this
How-To uses only numeric IP addresses.
However, if DNS resolution is available on your network, it is
possible to use host names in lieu of IP addresses in
configuring Cluster. Alternatively, you can use the
/etc/hosts file or your operating
system's equivalent for providing a means to do host lookup
if such is available.
Note
A common problem when trying to use host names for Cluster
nodes arises because of the way in which some operating
systems (including some Linux distributions) set up the
system's own host name in the /etc/hosts
during installation. Consider two machines with the host names
ndb1 and ndb2 , both in
the cluster network domain. Red Hat Linux
(including some derivatives such as CentOS and Fedora) places
the following entries in these machines'
/etc/hosts files:
# ndb1 /etc/hosts :
127.0.0.1 ndb1.cluster ndb1 localhost.localdomain localhost
# ndb2 /etc/hosts :
127.0.0.1 ndb2.cluster ndb2 localhost.localdomain localhost
SUSE Linux (including OpenSUSE) places these entries in the
machines' /etc/hosts files:
# ndb1 /etc/hosts :
127.0.0.1 localhost
127.0.0.2 ndb1.cluster ndb1
# ndb2 /etc/hosts :
127.0.0.1 localhost
127.0.0.2 ndb2.cluster ndb2
In both instances, ndb1 routes
ndb1.cluster to a loopback IP address, but
gets a public IP address from DNS for
ndb2.cluster , while ndb2
routes ndb2.cluster to a loopback address
and obtains a public address for
ndb1.cluster . The result is that each data
node connects to the management server, but cannot tell when
any other data nodes have connected, and so the data nodes
appear to hang while starting.
You should also be aware that you cannot mix
localhost and other host names or IP
addresses in config.ini . For these
reasons, the solution in such cases (other than to use IP
addresses for all
config.ini HostName
entries) is to remove the fully qualified host names from
/etc/hosts and use these in
config.ini for all cluster hosts.
Each host in our scenario is an Intel-based desktop PC running a
common, generic Linux distribution installed to disk in a
standard configuration, and running no unnecessary services. The
core OS with standard TCP/IP networking capabilities should be
sufficient. Also for the sake of simplicity, we also assume that
the file systems on all hosts are set up identically. In the
event that they are not, you will need to adapt these
instructions accordingly.
Standard 100 Mbps or 1 gigabit Ethernet cards are installed on
each machine, along with the proper drivers for the cards, and
that all four hosts are connected via a standard-issue Ethernet
networking appliance such as a switch. (All machines should use
network cards with the same throughout. That is, all four
machines in the cluster should have 100 Mbps cards
or all four machines should have 1 Gbps
cards.) MySQL Cluster will work in a 100 Mbps network; however,
gigabit Ethernet will provide better performance.
Note that MySQL Cluster is not intended for
use in a network for which throughput is less than 100 Mbps. For
this reason (among others), attempting to run a MySQL Cluster
over a public network such as the Internet is not likely to be
successful, and is not recommended.
For our sample data, we will use the world
database which is available for download from the MySQL Web
site. As this database takes up a relatively small amount of
space, we assume that each machine has 256MB RAM, which should
be sufficient for running the operating system, host NDB
process, and (for the data nodes) for storing the database.
Although we refer to a Linux operating system in this How-To, the
instructions and procedures that we provide here should be easily
adaptable to other supported operating systems. We also assume that
you already know how to perform a minimal installation and
configuration of the operating system with networking capability, or
that you are able to obtain assistance in this elsewhere if needed.
For information about MySQL Cluster hardware, software, and
networking requirements, see
Section 17.1.3, “MySQL Cluster Hardware, Software, and Networking Requirements”.
17.2.1. MySQL Cluster Multi-Computer Installation
Each MySQL Cluster host computer running an SQL node must have
installed on it a MySQL binary. For management nodes and data
nodes, it is not necessary to install the MySQL server binary, but
management nodes require the management server daemon
(ndb_mgmd) and data nodes require the data node
daemon (ndbd; in MySQL Cluster NDB 7.0 and
later, you can use ndbmtd instead). It is also
a good idea to install the management client
(ndb_mgm) on the management server host. This
section covers the steps necessary to install the correct binaries
for each type of Cluster node.
Sun Microsystems, Inc. provides precompiled binaries that support
Cluster. However, we also include information relating to
installing a MySQL Cluster after building MySQL from source. For
setting up a cluster using MySQL's binaries, the first step
in the installation process for each cluster host is to download
the latest MySQL Cluster NDB 6.2, MySQL Cluster NDB 6.3, or MySQL
Cluster NDB 7.0 binary archive
(mysql-cluster-gpl-6.2.19-linux-i686-glibc23.tar.gz ,
mysql-cluster-gpl-6.3.32-linux-i686-glibc23.tar.gz ,
or
mysql-cluster-gpl-7.0.12-linux-i686-glibc23.tar.gz ,
respectively) from the
MySQL Cluster downloads
area. We assume that you have placed this file in each
machine's /var/tmp directory. (If you do
require a custom binary, see
Section 2.3.3, “Installing from the Development Source Tree”.)
When compiling MySQL Cluster NDB 7.0 from source, no special
options are required for building multi-threaded data node
binaries. On Unix platforms, configuring the build with any of the
options --plugins=max ,
--plugins=max-no-innodb , or
--with-ndbcluster causes
ndbmtd to be built automatically; make
install places the ndbmtd binary in
the libexec directory along with
mysqld, ndbd, and
ndb_mgm.
On Windows, beginning with MySQL Cluster NDB 7.0.11, using
WITH_NDBCLUSTER_STORAGE_ENGINE with
configure.js causes
ndbmtd.exe to be built automatically, and to be
found in the bin directory of the archive
created by make_win_bin_dist.
Important
Currently, MySQL Cluster is not compatible with the
InnoDB Plugin . You must use the version of
InnoDB that is supplied with the
MySQL Server. You can build MySQL Cluster with
InnoDB storage engine support using
the --with-plugins=max or
--with-innodb option for
configure.
This is a known issue, which we are working to address in a
future MySQL Cluster release.
RPMs are also available for both 32-bit and 64-bit Linux
platforms. For a MySQL Cluster, three RPMs are required:
The Server RPM (for example,
MySQL-Cluster-gpl-server-6.2.19-0.sles10.i586.rpm ,
MySQL-Cluster-gpl-server-6.3.32-0.sles10.i586.rpm ,
or
MySQL-Cluster-gpl-server-7.0.12-0.sles10.i586.rpm ),
which supplies the core files needed to run a MySQL Server
with NDBCLUSTER storage engine
support (that is, as a MySQL Cluster SQL node).
If you do not have your own client application capable of
administering a MySQL server, you should also obtain and
install the Client RPM (for
example,
MySQL-Cluster-gpl-client-6.2.19-0.sles10.i586.rpm ,
MySQL-Cluster-gpl-client-6.3.32-0.sles10.i586.rpm ,
or
MySQL-Cluster-gpl-client-7.0.12-0.sles10.i586.rpm ).
The Cluster storage engine
RPM (for example,
MySQL-Cluster-gpl-storage-6.2.19-0.sles10.i586.rpm ,
MySQL-Cluster-gpl-storage-6.3.32-0.sles10.i586.rpm ,
or
MySQL-Cluster-gpl-storage-7.0.12-0.sles10.i586.rpm ),
which supplies the MySQL Cluster data node binary
(ndbd).
The Cluster storage engine management
RPM (for example,
MySQL-Cluster-gpl-management-6.2.19-0.sles10.i586.rpm ,
MySQL-Cluster-gpl-management-6.3.32-0.sles10.i586.rpm ,
or
MySQL-Cluster-gpl-management-7.0.12-0.sles10.i586.rpm )
which provides the MySQL Cluster management server binary
(ndb_mgmd).
In addition, you should also obtain the NDB
Cluster - Storage engine basic tools RPM (for example,
MySQL-Cluster-gpl-tools-6.2.19-0.sles10.i586.rpm ,
MySQL-Cluster-gpl-tools-6.3.32-0.sles10.i586.rpm ,
or
MySQL-Cluster-gpl-tools-7.0.12-0.sles10.i586.rpm ),
which supplies several useful applications for working with a
MySQL Cluster. The most important of these is the MySQL Cluster
management client (ndb_mgm). The
NDB Cluster - Storage engine extra
tools RPM (for example,
MySQL-Cluster-gpl-extra-6.2.19-0.sles10.i586.rpm ,
MySQL-Cluster-gpl-extra-6.3.32-0.sles10.i586.rpm ,
or
MySQL-Cluster-gpl-extra-7.0.12-0.sles10.i586.rpm )
contains some additional testing and monitoring programs, but is
not required to install a MySQL Cluster. (For more information
about these additional programs, see
Section 17.4, “MySQL Cluster Programs”.)
The MySQL Cluster version number in the RPM file names (shown here
as 6.2.19 ,
6.3.32 , or
7.0.12 ) can vary according to
the version which you are actually using. It is very
important that all of the Cluster RPMs to be installed have the
same version number. The glibc
version number (if present), and architecture designation (shown
here as i586 ) should be appropriate to the
machine on which the RPM is to be installed.
See Section 2.6.1, “Installing MySQL from RPM Packages on Linux”, for general information about
installing MySQL using RPMs supplied by Sun Microsystems, Inc.
After installing from RPM, you still need to configure the cluster
as discussed in Section 17.2.2, “MySQL Cluster Multi-Computer Configuration”.
Note
After completing the installation, do not yet start any of the
binaries. We show you how to do so following the configuration
of all nodes.
Data and SQL Node Installation — .tar.gz
Binary.
On each of the machines designated to host data or SQL nodes,
perform the following steps as the system
root user:
Check your /etc/passwd and
/etc/group files (or use whatever
tools are provided by your operating system for managing
users and groups) to see whether there is already a
mysql group and
mysql user on the system. Some OS
distributions create these as part of the operating system
installation process. If they are not already present,
create a new mysql user group, and then
add a mysql user to this group:
shell> groupadd mysql
shell> useradd -g mysql mysql
The syntax for useradd and
groupadd may differ slightly on
different versions of Unix, or they may have different
names such as adduser and
addgroup.
Change location to the directory containing the downloaded
file, unpack the archive, and create a symlink to the
mysql directory named
mysql . Note that the actual file and
directory names will vary according to the MySQL Cluster
version number.
shell> cd /var/tmp
shell> tar -C /usr/local -xzvf mysql-cluster-gpl-7.0.12-linux-i686-glibc23.tar.gz
shell> ln -s /usr/local/mysql-cluster-gpl-7.0.12-linux-i686-glibc23.tar.gz /usr/local/mysql
Change location to the mysql
directory and run the supplied script for creating the
system databases:
shell> cd mysql
shell> scripts/mysql_install_db --user=mysql
Set the necessary permissions for the MySQL server and
data directories:
shell> chown -R root .
shell> chown -R mysql data
shell> chgrp -R mysql .
Note that the data directory on each machine hosting a
data node is /usr/local/mysql/data .
This piece of information is essential when configuring
the management node. (See
Section 17.2.2, “MySQL Cluster Multi-Computer Configuration”.)
Copy the MySQL startup script to the appropriate
directory, make it executable, and set it to start when
the operating system is booted up:
shell> cp support-files/mysql.server /etc/rc.d/init.d/
shell> chmod +x /etc/rc.d/init.d/mysql.server
shell> chkconfig --add mysql.server
(The startup scripts directory may vary depending on your
operating system and version — for example, in some
Linux distributions, it is
/etc/init.d .)
Here we use Red Hat's chkconfig for
creating links to the startup scripts; use whatever means
is appropriate for this purpose on your operating system
and distribution, such as update-rc.d
on Debian.
Remember that the preceding steps must be repeated on each
machine where an SQL node is to reside.
SQL node installation — RPM files.
On each machine to be used for hosting a cluster SQL node,
install the Server RPM by
executing the following command as the system root user,
replacing the name shown for the RPM as necessary to match the
name of the RPM downloaded from the MySQL web site:
shell> rpm -Uhv MySQL-Cluster-gpl-server-7.0.12-0.sles10.i586.rpm
This installs the MySQL server binary
(mysqld) in the
/usr/sbin directory, as well as all needed
MySQL Server support files. It also installs the
mysql.server and
mysqld_safe startup scripts in
/usr/share/mysql and
/usr/bin , respectively. The RPM installer
should take care of general configuration issues (such as
creating the mysql user and group, if needed)
automatically.
Note
To administer the SQL node (MySQL server), you should also
install the Client RPM, as
shown here:
shell> rpm -Uhv MySQL-Cluster-gpl-client-7.0.12-0.sles10.i586.rpm
This installs the mysql client program.
SQL node installation — building from source.
If you compile MySQL with clustering support (for example, by
using the
BUILD/compile-platform_name -max
script appropriate to your platform), and perform the default
installation (using make install as the root
user), mysqld is placed in
/usr/local/mysql/bin . Follow the steps
given in Section 2.3, “MySQL Installation Using a Source Distribution” to make
mysqld ready for use. If you want to run
multiple SQL nodes, you can use a copy of the same
mysqld executable and its associated support
files on several machines. The easiest way to do this is to copy
the entire /usr/local/mysql directory and
all directories and files contained within it to the other SQL
node host or hosts, then repeat the steps from
Section 2.3, “MySQL Installation Using a Source Distribution” on each machine. If you
configure the build with a nondefault
--prefix , you need to adjust
the directory accordingly.
Data node installation — RPM Files.
On a computer that is to host a cluster data node it is
necessary to install only the NDB Cluster
- Storage engine RPM. To do so, copy this RPM to the
data node host, and run the following command as the system root
user, replacing the name shown for the RPM as necessary to match
that of the RPM downloaded from the MySQL web site:
shell> rpm -Uhv MySQL-Cluster-gpl-storage-7.0.12-0.sles10.i586.rpm
The previous command installs the MySQL Cluster data node binary
(ndbd) in the /usr/sbin
directory.
Data node installation — building from source.
The only executable required on a data node host is
ndbd or (in MySQL Cluster NDB 7.0 and later)
ndbmtd (mysqld, for
example, does not have to be present on the host machine). By
default when doing a source build, this file is placed in the
directory /usr/local/mysql/libexec . For
installing on multiple data node hosts, only
ndbd need be copied to the other host machine
or machines. (This assumes that all data node hosts use the same
architecture and operating system; otherwise you may need to
compile separately for each different platform.)
ndbd need not be in any particular location
on the host's file system, as long as the location is known.
Note
ndbmtd was not built on Windows prior to
MySQL Cluster NDB 7.0.11.
Management node installation — .tar.gz binary.
Installation of the management node does not require the
mysqld binary. Only the MySQL Cluster
management server (ndb_mgmd) is required; you
most likely want to install the management client
(ndb_mgm) as well. Both of these binaries
also be found in the .tar.gz archive.
Again, we assume that you have placed this archive in
/var/tmp .
As system root (that is, after using
sudo, su root, or your
system's equivalent for temporarily assuming the system
administrator account's privileges), perform the following steps
to install ndb_mgmd and
ndb_mgm on the Cluster management node host:
Change location to the /var/tmp
directory, and extract the ndb_mgm and
ndb_mgmd from the archive into a suitable
directory such as /usr/local/bin :
shell> cd /var/tmp
shell> tar -zxvf mysql-5.1.41-ndb-7.0.12-linux-i686-glibc23.tar.gz
shell> cd mysql-5.1.41-ndb-7.0.12-linux-i686-glibc23
shell> cp bin/ndb_mgm* /usr/local/bin
(You can safely delete the directory created by unpacking
the downloaded archive, and the files it contains, from
/var/tmp once
ndb_mgm and ndb_mgmd
have been copied to the executables directory.)
Change location to the directory into which you copied the
files, and then make both of them executable:
shell> cd /usr/local/bin
shell> chmod +x ndb_mgm*
Management node installation — RPM file.
To install the MySQL Cluster management server, it is necessary
only to use the NDB Cluster - Storage
engine management RPM. Copy this RPM to the computer
intended to host the management node, and then install it by
running the following command as the system root user (replace
the name shown for the RPM as necessary to match that of the
Storage engine management RPM
downloaded from the MySQL web site):
shell> rpm -Uhv MySQL-Cluster-gpl-management-7.0.12-0.sles10.i586.rpm
This installs the management server binary
(ndb_mgmd) to the
/usr/sbin directory.
You should also install the NDB
management client, which is supplied by the
Storage engine basic tools RPM.
Copy this RPM to the same computer as the management node, and
then install it by running the following command as the system
root user (again, replace the name shown for the RPM as necessary
to match that of the Storage engine basic
tools RPM downloaded from the MySQL web site):
shell> rpm -Uhv MySQL-Cluster-gpl-tools-7.0.12-0.sles10.i586.rpm
The Storage engine basic tools
RPM installs the MySQL Cluster management client
(ndb_mgm) to the /usr/bin
directory.
Note
You can also install the Cluster storage
engine extra tools RPM, if you wish, as shown here:
shell> rpm -Uhv MySQL-Cluster-gpl-extra-7.0.12-0.sles10.i586.rpm
You may find the extra tools useful; however the
Cluster storage engine extra
tools RPM is not required to
install a working MySQL Cluster.
Management node installation — building from source.
When building from source and running the default make
install, the management server binary
(ndb_mgmd) is placed in
/usr/local/mysql/libexec , while the
management client binary (ndb_mgm) can be
found in /usr/local/mysql/bin . Only
ndb_mgmd is required to be present on a
management node host; however, it is also a good idea to have
ndb_mgm present on the same host machine.
Neither of these executables requires a specific location on the
host machine's file system.
In Section 17.2.2, “MySQL Cluster Multi-Computer Configuration”, we create
configuration files for all of the nodes in our example MySQL
Cluster.
MySQL Cluster on Windows (alpha).
In MySQL Cluster NDB 7.0, experimental support is added for
Microsoft Windows platforms. To compile MySQL Cluster from
source on Windows, you must configure the build using the
WITH_NDBCLUSTER_STORAGE_ENGINE option before
creating the Visual Studio project files. After running
make_win_bin_dist, the MySQL Cluster binaries
can be found in the bin directory of the
resulting archive. For more information, see
Section 2.5.10, “Installing MySQL from Source on Windows”.
17.2.2. MySQL Cluster Multi-Computer Configuration
For our four-node, four-host MySQL Cluster, it is necessary to
write four configuration files, one per node host.
Each data node or SQL node requires a
my.cnf file that provides two pieces of
information: a connectstring that tells
the node where to find the management node, and a line telling
the MySQL server on this host (the machine hosting the data
node) to enable the NDBCLUSTER
storage engine.
For more information on connectstrings, see
Section 17.3.2.3, “The MySQL Cluster Connectstring”.
The management node needs a config.ini
file telling it how many replicas to maintain, how much memory
to allocate for data and indexes on each data node, where to
find the data nodes, where to save data to disk on each data
node, and where to find any SQL nodes.
Configuring the Storage and SQL
Nodes
The my.cnf file needed for the data nodes is
fairly simple. The configuration file should be located in the
/etc directory and can be edited using any
text editor. (Create the file if it does not exist.) For example:
shell> vi /etc/my.cnf
Note
We show vi being used here to create the
file, but any text editor should work just as well.
For each data node and SQL node in our example setup,
my.cnf should look like this:
# Options for mysqld process:
[mysqld]
ndbcluster # run NDB storage engine
ndb-connectstring=192.168.0.10 # location of management server
# Options for ndbd process:
[mysql_cluster]
ndb-connectstring=192.168.0.10 # location of management server
After entering the preceding information, save this file and exit
the text editor. Do this for the machines hosting data node
“A”, data node “B”, and the SQL node.
Important
Once you have started a mysqld process with
the NDBCLUSTER and
ndb-connectstring parameters in the
[mysqld] in the my.cnf
file as shown previously, you cannot execute any
CREATE TABLE or
ALTER TABLE statements without
having actually started the cluster. Otherwise, these statements
will fail with an error. This is by design.
Configuring the management node.
The first step in configuring the management node is to create
the directory in which the configuration file can be found and
then to create the file itself. For example (running as
root ):
shell> mkdir /var/lib/mysql-cluster
shell> cd /var/lib/mysql-cluster
shell> vi config.ini
For our representative setup, the config.ini
file should read as follows:
# Options affecting ndbd processes on all data nodes:
[ndbd default]
NoOfReplicas=2 # Number of replicas
DataMemory=80M # How much memory to allocate for data storage
IndexMemory=18M # How much memory to allocate for index storage
# For DataMemory and IndexMemory, we have used the
# default values. Since the "world" database takes up
# only about 500KB, this should be more than enough for
# this example Cluster setup.
# TCP/IP options:
[tcp default]
portnumber=2202 # This the default; however, you can use any port that is free
# for all the hosts in the cluster
# Note: It is recommended that you do not specify the port
# number at all and allow the default value to be used instead
# Management process options:
[ndb_mgmd]
hostname=192.168.0.10 # Hostname or IP address of management node
datadir=/var/lib/mysql-cluster # Directory for management node log files
# Options for data node "A":
[ndbd]
# (one [ndbd] section per data node)
hostname=192.168.0.30 # Hostname or IP address
datadir=/usr/local/mysql/data # Directory for this data node's data files
# Options for data node "B":
[ndbd]
hostname=192.168.0.40 # Hostname or IP address
datadir=/usr/local/mysql/data # Directory for this data node's data files
# SQL node options:
[mysqld]
hostname=192.168.0.20 # Hostname or IP address
# (additional mysqld connections can be
# specified for this node for various
# purposes such as running ndb_restore)
After all the configuration files have been created and these
minimal options have been specified, you are ready to proceed with
starting the cluster and verifying that all processes are running.
We discuss how this is done in
Section 17.2.3, “Initial Startup of MySQL Cluster”.
For more detailed information about the available MySQL Cluster
configuration parameters and their uses, see
Section 17.3.2, “MySQL Cluster Configuration Files”, and
Section 17.3, “MySQL Cluster Configuration”. For configuration
of MySQL Cluster as relates to making backups, see
Section 17.5.3.3, “Configuration for MySQL Cluster Backups”.
Note
The default port for Cluster management nodes is 1186; the
default port for data nodes is 2202. However, the cluster can
automatically allocate ports for data nodes from those that are
already free.
17.2.3. Initial Startup of MySQL Cluster
Starting the cluster is not very difficult after it has been
configured. Each cluster node process must be started separately,
and on the host where it resides. The management node should be
started first, followed by the data nodes, and then finally by any
SQL nodes:
On the management host, issue the following command from the
system shell to start the management node process:
shell> ndb_mgmd -f /var/lib/mysql-cluster/config.ini
On each of the data node hosts, run this command to start the
ndbd process:
shell> ndbd
If you used RPM files to install MySQL on the cluster host
where the SQL node is to reside, you can (and should) use the
supplied startup script to start the MySQL server process on
the SQL node.
If all has gone well, and the cluster has been set up correctly,
the cluster should now be operational. You can test this by
invoking the ndb_mgm management node client.
The output should look like that shown here, although you might
see some slight differences in the output depending upon the exact
version of MySQL that you are using:
shell> ndb_mgm
-- NDB Cluster -- Management Client --
ndb_mgm> SHOW
Connected to Management Server at: localhost:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=2 @192.168.0.30 (Version: 5.1.41-ndb-6.3.32, Nodegroup: 0, Master)
id=3 @192.168.0.40 (Version: 5.1.41-ndb-6.3.32, Nodegroup: 0)
[ndb_mgmd(MGM)] 1 node(s)
id=1 @192.168.0.10 (Version: 5.1.41-ndb-6.3.32)
[mysqld(API)] 1 node(s)
id=4 @192.168.0.20 (Version: 5.1.41-ndb-6.3.32)
The SQL node is referenced here as
[mysqld(API)] , which reflects the fact that the
mysqld process is acting as a MySQL Cluster API
node.
Note
The IP address shown for a given MySQL Cluster SQL or other API
node in the output of SHOW
is the address used by the SQL or API node to connect to the
cluster data nodes, and not to any management node.
You should now be ready to work with databases, tables, and data
in MySQL Cluster. See
Section 17.2.4, “Loading Sample Data into MySQL Cluster and Performing Queries”, for a
brief discussion.
17.2.4. Loading Sample Data into MySQL Cluster and Performing Queries
Working with data in MySQL Cluster is not much different from
doing so in MySQL without Cluster. There are two points to keep in
mind:
For a table to be replicated in the cluster, it must use the
NDBCLUSTER storage engine. To
specify this, use the ENGINE=NDBCLUSTER or
ENGINE=NDB option when creating the table:
CREATE TABLE tbl_name (col_name column_definitions ) ENGINE=NDBCLUSTER;
Alternatively, for an existing table that uses a different
storage engine, use ALTER TABLE
to change the table to use
NDBCLUSTER :
ALTER TABLE tbl_name ENGINE=NDBCLUSTER;
Each NDBCLUSTER table
must have a primary key. If no primary
key is defined by the user when a table is created, the
NDBCLUSTER storage engine
automatically generates a hidden one.
Note
This hidden key takes up space just as does any other table
index. It is not uncommon to encounter problems due to
insufficient memory for accommodating these automatically
created indexes.)
If you are importing tables from an existing database using the
output of mysqldump, you can open the SQL
script in a text editor and add the ENGINE
option to any table creation statements, or replace any existing
ENGINE (or TYPE ) options.
Suppose that you have the world sample database
on another MySQL server that does not support MySQL Cluster, and
you want to export the City table:
shell> mysqldump --add-drop-table world City > city_table.sql
The resulting city_table.sql file will
contain this table creation statement (and the
INSERT statements necessary to
import the table data):
DROP TABLE IF EXISTS `City`;
CREATE TABLE `City` (
`ID` int(11) NOT NULL auto_increment,
`Name` char(35) NOT NULL default '',
`CountryCode` char(3) NOT NULL default '',
`District` char(20) NOT NULL default '',
`Population` int(11) NOT NULL default '0',
PRIMARY KEY (`ID`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1;
INSERT INTO `City` VALUES (1,'Kabul','AFG','Kabol',1780000);
INSERT INTO `City` VALUES (2,'Qandahar','AFG','Qandahar',237500);
INSERT INTO `City` VALUES (3,'Herat','AFG','Herat',186800);
(remaining INSERT statements omitted)
You need to make sure that MySQL uses the
NDBCLUSTER storage engine for this
table. There are two ways that this can be accomplished. One of
these is to modify the table definition
before importing it into the Cluster
database. Using the City table as an example,
modify the ENGINE option of the definition as
follows:
DROP TABLE IF EXISTS `City`;
CREATE TABLE `City` (
`ID` int(11) NOT NULL auto_increment,
`Name` char(35) NOT NULL default '',
`CountryCode` char(3) NOT NULL default '',
`District` char(20) NOT NULL default '',
`Population` int(11) NOT NULL default '0',
PRIMARY KEY (`ID`)
) ENGINE=NDBCLUSTER DEFAULT CHARSET=latin1;
INSERT INTO `City` VALUES (1,'Kabul','AFG','Kabol',1780000);
INSERT INTO `City` VALUES (2,'Qandahar','AFG','Qandahar',237500);
INSERT INTO `City` VALUES (3,'Herat','AFG','Herat',186800);
(remaining INSERT statements omitted)
This must be done for the definition of each table that is to be
part of the clustered database. The easiest way to accomplish this
is to do a search-and-replace on the file that contains the
definitions and replace all instances of
TYPE=engine_name or
ENGINE=engine_name
with ENGINE=NDBCLUSTER . If you do not want to
modify the file, you can use the unmodified file to create the
tables, and then use ALTER TABLE to
change their storage engine. The particulars are given later in
this section.
Assuming that you have already created a database named
world on the SQL node of the cluster, you can
then use the mysql command-line client to read
city_table.sql , and create and populate the
corresponding table in the usual manner:
shell> mysql world < city_table.sql
It is very important to keep in mind that the preceding command
must be executed on the host where the SQL node is running (in
this case, on the machine with the IP address
192.168.0.20 ).
To create a copy of the entire world database
on the SQL node, use mysqldump on the
noncluster server to export the database to a file named
world.sql ; for example, in the
/tmp directory. Then modify the table
definitions as just described and import the file into the SQL
node of the cluster like this:
shell> mysql world < /tmp/world.sql
If you save the file to a different location, adjust the preceding
instructions accordingly.
Running SELECT queries on the SQL
node is no different from running them on any other instance of a
MySQL server. To run queries from the command line, you first need
to log in to the MySQL Monitor in the usual way (specify the
root password at the Enter
password: prompt):
shell> mysql -u root -p
Enter password:
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 1 to server version: 5.1.41-ndb-6.2.19
Type 'help;' or '\h' for help. Type '\c' to clear the buffer.
mysql>
We simply use the MySQL server's root
account and assume that you have followed the standard security
precautions for installing a MySQL server, including setting a
strong root password. For more information, see
Section 2.13.2, “Securing the Initial MySQL Accounts”.
It is worth taking into account that Cluster nodes do
not make use of the MySQL privilege system
when accessing one another. Setting or changing MySQL user
accounts (including the root account) effects
only applications that access the SQL node, not interaction
between nodes. See
Section 17.5.9.2, “MySQL Cluster and MySQL Privileges”, for
more information.
If you did not modify the ENGINE clauses in the
table definitions prior to importing the SQL script, you should
run the following statements at this point:
mysql> USE world;
mysql> ALTER TABLE City ENGINE=NDBCLUSTER;
mysql> ALTER TABLE Country ENGINE=NDBCLUSTER;
mysql> ALTER TABLE CountryLanguage ENGINE=NDBCLUSTER;
Selecting a database and running a SELECT query
against a table in that database is also accomplished in the usual
manner, as is exiting the MySQL Monitor:
mysql> USE world;
mysql> SELECT Name, Population FROM City ORDER BY Population DESC LIMIT 5;
+-----------+------------+
| Name | Population |
+-----------+------------+
| Bombay | 10500000 |
| Seoul | 9981619 |
| S?o Paulo | 9968485 |
| Shanghai | 9696300 |
| Jakarta | 9604900 |
+-----------+------------+
5 rows in set (0.34 sec)
mysql> \q
Bye
shell>
Applications that use MySQL can employ standard APIs to access
NDB tables. It is important to
remember that your application must access the SQL node, and not
the management or data nodes. This brief example shows how we
might execute the SELECT statement
just shown by using the PHP 5.X mysqli
extension running on a Web server elsewhere on the network:
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
"http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<meta http-equiv="Content-Type"
content="text/html; charset=iso-8859-1">
<title>SIMPLE mysqli SELECT</title>
</head>
<body>
<?php
# connect to SQL node:
$link = new mysqli('192.168.0.20', 'root', 'root_password ', 'world');
# parameters for mysqli constructor are:
# host, user, password, database
if( mysqli_connect_errno() )
die("Connect failed: " . mysqli_connect_error());
$query = "SELECT Name, Population
FROM City
ORDER BY Population DESC
LIMIT 5";
# if no errors...
if( $result = $link->query($query) )
{
?>
<table border="1" width="40%" cellpadding="4" cellspacing ="1">
<tbody>
<tr>
<th width="10%">City</th>
<th>Population</th>
</tr>
<?
# then display the results...
while($row = $result->fetch_object())
printf("<tr>\n <td align=\"center\">%s</td><td>%d</td>\n</tr>\n",
$row->Name, $row->Population);
?>
</tbody
</table>
<?
# ...and verify the number of rows that were retrieved
printf("<p>Affected rows: %d</p>\n", $link->affected_rows);
}
else
# otherwise, tell us what went wrong
echo mysqli_error();
# free the result set and the mysqli connection object
$result->close();
$link->close();
?>
</body>
</html>
We assume that the process running on the Web server can reach the
IP address of the SQL node.
In a similar fashion, you can use the MySQL C API, Perl-DBI,
Python-mysql, or MySQL Connectors to perform the tasks of data
definition and manipulation just as you would normally with MySQL.
17.2.5. Safe Shutdown and Restart of MySQL Cluster
To shut down the cluster, enter the following command in a shell
on the machine hosting the management node:
shell> ndb_mgm -e shutdown
The -e option here is used to pass a command to
the ndb_mgm client from the shell. (See
Section 17.4.23, “Options Common to MySQL Cluster Programs”, for more
information about this option.) The command causes the
ndb_mgm, ndb_mgmd, and any
ndbd processes to terminate gracefully. Any SQL
nodes can be terminated using mysqladmin
shutdown and other means.
To restart the cluster, run these commands:
On the management host (192.168.0.10 in our
example setup):
shell> ndb_mgmd -f /var/lib/mysql-cluster/config.ini
On each of the data node hosts
(192.168.0.30 and
192.168.0.40 ):
shell> ndbd
On the SQL host (192.168.0.20 ):
shell> mysqld_safe &
In a production setting, it is usually not desirable to shut down
the cluster completely. In many cases, even when making
configuration changes, or performing upgrades to the cluster
hardware or software (or both), which require shutting down
individual host machines, it is possible to do so without shutting
down the cluster as a whole by performing a rolling
restart of the cluster. For more information about
doing this, see Section 17.2.6.1, “Performing a Rolling Restart of a MySQL Cluster”.
17.2.6. Upgrading and Downgrading MySQL Cluster
This portion of the MySQL Cluster chapter covers upgrading and
downgrading a MySQL Cluster from one MySQL release to another. It
discusses different types of Cluster upgrades and downgrades, and
provides a Cluster upgrade/downgrade compatibility matrix (see
Section 17.2.6.2, “MySQL Cluster 5.1 and MySQL Cluster NDB 6.x/7.x Upgrade and Downgrade
Compatibility”).
You are expected already to be familiar with installing and
configuring a MySQL Cluster prior to attempting an upgrade or
downgrade. See Section 17.3, “MySQL Cluster Configuration”.
For more information about upgrading or downgrading between MySQL
Cluster NDB releases, or between MySQL Cluster NDB releases and
mainline MySQL releases, see the changelogs relating to the
applicable MySQL Cluster versions.
This section remains in development, and continues to be updated
and expanded.
17.2.6.1. Performing a Rolling Restart of a MySQL Cluster
This section discusses how to perform a rolling
restart of a MySQL Cluster installation, so called
because it involves stopping and starting (or restarting) each
node in turn, so that the cluster itself remains operational.
This is often done as part of a rolling
upgrade or rolling downgrade,
where high availability of the cluster is mandatory and no
downtime of the cluster as a whole is permissible. Where we
refer to upgrades, the information provided here also generally
applies to downgrades as well.
There are a number of reasons why a rolling restart might be
desirable:
Cluster configuration change.
To make a change in the cluster's configuration, such
as adding an SQL node to the cluster, or setting a
configuration parameter to a new value.
Cluster software upgrade/downgrade.
To upgrade the cluster to a newer version of the MySQL
Cluster software (or to downgrade it to an older version).
This is usually referred to as a “rolling
upgrade” (or “rolling downgrade”, when
reverting to an older version of MySQL Cluster).
Change on node host.
To make changes in the hardware or operating system on
which one or more cluster nodes are running.
Cluster reset.
To reset the cluster because it has reached an undesirable
state. In such cases it is often desirable to reload the
data and metadata of one or more data nodes. This can be
done 1 of 3 ways:
Freeing of resources.
To allow memory allocated to a table by successive
INSERT and
DELETE operations to be
freed for re-use by other MySQL Cluster tables.
The process for performing a rolling restart may be generalized
as follows:
Stop all cluster management nodes
(ndb_mgmd processes), reconfigure them,
then restart them.
Stop, reconfigure, then restart each cluster data node
(ndbd process) in turn.
Stop, reconfigure, then restart each cluster SQL node
(mysqld process) in turn.
The specifics for implementing a particular rolling upgrade
depend upon the actual changes being made. A more detailed view
of the process is presented here:
In the previous diagram, the
Stop and
Start steps indicate that the
process must be stopped completely using a shell command (such
as kill on most Unix systems) or the
management client STOP command, then started
again from a system shell by invoking the
ndbd or ndb_mgmd
executable as appropriate.
Restart indicates the process
may be restarted using the ndb_mgm management
client RESTART command.
Prior to MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB 7.0.10.
When performing an upgrade or downgrade of the cluster
software, you must upgrade or downgrade
the management nodes first, then the data
nodes, and finally the SQL nodes. Doing so in any other order
may leave the cluster in an unusable state.
MySQL Cluster NDB 6.3.29 and later; MySQL Cluster NDB 7.0.10 and later.
MySQL Cluster supports a more flexible order for upgrading the
cluster nodes. When upgrading a cluster running MySQL Cluster
NDB 6.3.29 or later, or a cluster that is running MySQL
Cluster NDB 7.0.10 or later, you may upgrade API nodes
(including SQL nodes) before upgrading the management nodes,
data nodes, or both. In other words, you are permitted to
upgrade the API and SQL nodes in any order. This is subject to
the following provisions:
This functionality is intended for use as part of an online
upgrade only. A mix of node binaries from different MySQL
Cluster releases is neither intended nor supported for
constant, long-term use in a production setting.
All management nodes must be upgraded before any data nodes
are upgraded. This remains true regardless of the order in
which you upgrade the cluster's API and SQL nodes.
For MySQL Cluster NDB 6.3, the ability to upgrade API nodes
in any order relative to upgading management nodes and data
nodes is supported only for MySQL Cluster NDB 6.3.29 and
later; for MySQL Cluster NDB 7.0, it supported only for
MySQL Cluster NDB 7.0.10 and later. This means that, if you
are upgrading from a MySQL Cluster NDB 6.3 release to a
MySQL Cluster NDB 7.0 release, the “old”
NDB engine version must be
6.3.29 or later, and the “new”
NDB engine version must be
7.0.10 or later.
When upgrading the cluster from a MySQL Cluster NDB 6.3
release to a MySQL Cluster NDB 7.0 release: Once you have
started upgraded the API nodes, you should not perform DDL
operations until all management nodes and data nodes have
been upgraded. DML operations should be unaffected, and can
continue while the upgrade is in progress.
However, it is possible to perform DDL from an
“old” (NDB 6.3 version) API node as long as the
master data node is also running the “old”
version of MySQL Cluster. You should keep in mind that a
data node restart could result in the master node running a
“new” (NDB 7.0 version) binary while one or
more data nodes are still using the “old” (NDB
6.3) version; in this situation, no DDL can be performed
from any API node, because the master
data node is no longer using an NDB 6.3 binary, but the
cluster still contains nodes which are not yet using NDB
7.0. For this reason, we recommend that you avoid performing
DDL at any time while the upgrade is in progress.
Features specific to the “new” version must not
be used until all management nodes and data nodes have been
upgraded.
This also applies to any MySQL Server version change that
may apply, in addition to the NDB engine version change, so
do not forget to take this into account when planning the
upgrade. (This is true for online upgrades of MySQL Cluster
in general.)
See also Bug#48528 and Bug#49163.
17.2.6.2. MySQL Cluster 5.1 and MySQL Cluster NDB 6.x/7.x Upgrade and Downgrade
Compatibility
This section provides information about MySQL Cluster software
and table file compatibility between MySQL 5.1 and
MySQL Cluster NDB 6.x releases with regard to performing
upgrades and downgrades.
Important
Only compatibility between MySQL versions with regard to
NDBCLUSTER is taken into account
in this section, and there are likely other issues to be
considered. As with any other MySQL software upgrade
or downgrade, you are strongly encouraged to review the
relevant portions of the MySQL Manual for the MySQL versions
from which and to which you intend to migrate, before
attempting an upgrade or downgrade of the MySQL Cluster
software. See Section 2.4.1, “Upgrading MySQL”.
The following table shows Cluster upgrade and downgrade
compatibility between different releases of MySQL
5.1:
Notes — MySQL 5.1.
MySQL 5.1.3 was the first public release in this series.
Direct upgrades or downgrades between MySQL Cluster 5.0
and 5.1 are not supported; you must dump all
NDBCLUSTER tables using
mysqldump, install the new version of
the software, and then reload the tables from the dump.
You cannot downgrade a MySQL 5.1.6 or later Cluster
using Disk Data tables to MySQL 5.1.5 or earlier unless
you convert all such tables to in-memory Cluster tables
first.
MySQL 5.1.8, MySQL 5.1.10, and MySQL 5.1.13 were not
released.
Online cluster upgrades and downgrades between MySQL
5.1.11 (or an earlier version) and 5.1.12 (or a later
version) are not possible due to major changes in the
cluster file system. In such cases, you must perform a
backup or dump, upgrade (or downgrade) the software,
start each data node with --initial ,
and then restore from the backup or dump. You can use
NDB backup/restore or
mysqldump for this purpose.
Online downgrades from MySQL 5.1.14 or later to versions
previous to 5.1.14 are not supported due to incompatible
changes in the cluster system tables.
Online upgrades from MySQL 5.1.17 and earlier to 5.1.18
and later MySQL 5.1.x releases are not supported for
clusters using replication due to incompatible changes
in the mysql.ndb_apply_status table.
(Online upgrades from MySQL 5.1 to MySQL Cluster NDB 6.2
and later are not supported, as discussed elsewhere in
this section.) However, it should not be necessary to
shut down the cluster entirely, if you follow this
modified rolling restart procedure:
Stop the management server, update the management
server software, then start the management server
again. For multiple management servers, repeat this
step for each management server in turn.
For each data node in turn: Stop the data node,
update the data node daemon (in MySQL Cluster NDB
7.0 and later, this can be either
ndbd or
ndbmtd) with the new version,
then restart the data node. It should not be
necessary to use --initial when
restarting any of the data nodes after updating the
software.
Stop all SQL nodes. Upgrade the
existing MySQL server installations to the new
version on all SQL nodes, then restart them. It is
not necessary to start them one at a time after
upgrading the MySQL server software, but
there must be a time when none of them is
running before starting any of them again using the
5.1.18 (or later)
mysqld. Otherwise
— due to the fact that
mysql.ndb_apply_status uses the
NDB storage engine and
is thus shared between all SQL nodes — there
may be conflicts between the old and new versions of
the table on different SQL nodes.
You can find more information about the changes to
ndb_apply_status in
Section 17.6.4, “MySQL Cluster Replication Schema and Tables”.
As with any other MySQL Cluster version upgrade, you
should also update the MySQL Cluster management client
(ndb_mgm) and other MySQL Cluster
client programs such as ndb_config
and ndb_error_reporter; however, this
does not have to be done in any particular order.
The internal specifications for columns in
NDBCLUSTER tables changed
in MySQL 5.1.18 to allow compatibility with later MySQL
Cluster releases that allow online adding and dropping
of columns. This change is not
backward-compatible with earlier MySQL
versions.
In order to make tables created in MySQL 5.1.17 and
earlier compatible with online adding and dropping of
columns (available beginning with beginning with MySQL
Cluster NDB 6.2.5 and MySQL Cluster NDB 6.3.3 —
see Section 12.1.7, “ALTER TABLE Syntax”, for more
information), it is necessary to force MySQL 5.1.18 and
later to convert the tables to the new format by
following this procedure:
Back up all
NDBCLUSTER tables.
Upgrade the MySQL Cluster software on all data,
management, and SQL nodes.
Shut down the cluster completely (this includes
all data, management, and API or SQL nodes).
Restart the cluster, starting all data nodes with
the --initial option (to clear
and rebuild the data node file systems).
Restore the
NDBCLUSTER tables
from backup.
This is not necessary for
NDBCLUSTER tables created
in MySQL 5.1.18 and later; such tables will
automatically be compatible with online adding and
dropping of columns (as implemented beginning with MySQL
Cluster NDB 6.2.5 and MySQL Cluster NDB 6.3.2).
In order to minimise possible later difficulties, it is
strongly advised that the procedure outlined above be
followed as soon as possible after to upgrading from
MySQL 5.1.17 or earlier to MySQL 5.1.18 or later.
Information about how this change effects users of MySQL
Cluster NDB 6.x/7.0 is provided later in this section.
MySQL Cluster is not supported in standard MySQL 5.1
releases, beginning with MySQL 5.1.25. If you are using
MySQL Cluster in a standard MySQL 5.1 release, you
should upgrade to the most recent MySQL Cluster NDB 6.2
or 6.3 release.
The following table shows Cluster upgrade and downgrade
compatibility between different releases of MySQL Cluster NDB
6.x/7.x:
Notes — MySQL Cluster NDB 6.x/7.x.
MySQL Cluster NDB 6.1 is no longer in production; if you
are still using a MySQL Cluster NDB 6.1 release, you
should upgrade to the most recent MySQL Cluster NDB 6.2
or 6.3 as soon as possible.
It is not possible to upgrade from MySQL Cluster NDB
6.1.2 (or an older 6.1 release) directly to 6.1.4 or a
newer NDB 6.1 release, or to downgrade from 6.1.4 (or a
newer 6.1 release) directly to 6.1.2 or an older NDB 6.1
release; in either case, you must upgrade or downgrade
to MySQL Cluster NDB 6.1.3 first.
It is not possible to perform an online downgrade from
MySQL Cluster NDB 6.1.8 (or a newer 6.1 release) to
MySQL Cluster NDB 6.1.7 (or an older 6.1 release).
MySQL Cluster NDB 6.1.6 and 6.1.18 were not released.
It is not possible to perform an online upgrade or
downgrade between MySQL Cluster NDB 6.2 and any previous
release series (including mainline MySQL 5.1 and MySQL
Cluster NDB 6.1); it is necessary to perform a dump and
reload. However, it should be possible to perform online
upgrades or downgrades between any MySQL Cluster NDB 6.2
release and any MySQL Cluster NDB 6.3 release up to and
including 6.3.7.
The internal specifications for columns in
NDB tables changed in MySQL
Cluster NDB 6.1.17 and 6.2.1 to allow compatibility with
future MySQL Cluster releases that are expected to
implement online adding and dropping of columns. This
change is not backward-compatible with earlier MySQL or
MySQL Cluster NDB 6.x versions.
In order to make tables created in earlier versions
compatible with online adding and dropping of columns in
later versions, it is necessary to force MySQL Cluster
to convert the tables to the new format by following
this procedure following an upgrade:
Upgrade the MySQL Cluster software on all data,
management, and SQL nodes
Back up all NDB
tables
Shut down the cluster (all data, management, and
SQL nodes)
Restart the cluster, starting all data nodes with
the --initial option (to clear
and rebuild the data node file systems)
Restore the tables from backup
In order to minimise possible later difficulties, it is
strongly advised that the procedure outlined above be
followed as soon as possible after to upgrading between
the versions indicated. The procedure is
not necessary for
NDBCLUSTER tables created
in any of the following versions:
MySQL Cluster NDB 6.1.8 or a later MySQL Cluster
NDB 6.1 release
MySQL Cluster 6.2.1 or a later MySQL Cluster NDB
6.2 release
Any MySQL Cluster NDB 6.3 release
Tables created in the listed versions (or later ones, as
indicated) are already compatible with online adding and
dropping of columns (as implemented beginning with MySQL
Cluster NDB 6.2.5 and MySQL Cluster NDB 6.3.2).
It was not possible to perform an online upgrade between
any MySQL Cluster NDB 6.2 release and MySQL Cluster NDB
6.3.8 and later MySQL Cluster 6.3 releases. This issue
was fixed in MySQL Cluster NDB 6.3.21. (Bug#41435)
Online downgrades between MySQL Cluster NDB 6.2.5 and
earlier releases are not supported.
Online downgrades between MySQL Cluster NDB 6.3.8 and
earlier releases are not supported.
Online upgrades from any MySQL Cluster NDB 7.0 release
up to and including MySQL Cluster NDB 7.0.4 (as well as
all early releases numbered NDB 6.4.x) to MySQL Cluster
NDB 7.0.5 or later are not possible. Upgrades to MySQL
Cluster NDB 7.0.6 or later from MySQL Cluster NDB 6.3.8
or a later MySQL Cluster NDB 6.3 release, or from MySQL
Cluster NDB 7.0.5 or later, are supported. (Bug#44294)
When upgrading online from a MySQL Cluster NDB 6.3
release to a MySQL Cluster NDB 7.0 release, you should
not try to upgrade the data nodes from
ndbd to ndbmtd at
the same time. Instead, perform the upgrade using the
new ndbd executable (from the MySQL
Cluster NDB 7.0.x distribution to which you are
upgrading) to replace the one in use on the data nodes.
Once the version upgrade is complete, you can perform a
second (online) upgrade to replace the data node
executables with ndbmtd from the
MySQL Cluster NDB 7.0.x distribution.
In MySQL Cluster NDB 7.0.4, the default values for a
number of MySQL Cluster configuration parameters
relating to memory usage and buffering changed (see
Section 17.7.2.13, “Changes in MySQL Cluster NDB 7.0.4 (5.1.32-ndb-7.0.4) (18 March 2008)”,
for a list of the parameters whose defaults changed).
For this reason, you may encounter issues if you try to
use a configuration that does not explicitly define each
of these buffers (because it was developed for a
previous version of MySQL Cluster, SendBufferMemory and
ReceiveBufferMemory in particular.
Prior to MySQL Cluster NDB 7.0.7, DML statements failed
if executed while performing an online upgrade from a
MySQL Cluster NDB 6.3 release. (Bug#45917)
Following an upgrade from any MySQL Cluster NDB 6.3.x
release to MySQL Cluster NDB 7.0.6, DDL and backup
operations failed. This issue was resolved in MySQL
Cluster NDB 7.0.7. (Bug#46494, Bug#46563)
In some cases, there could be problems with online
upgrades from MySQL Cluster NDB 6.3 releases to MySQL
Cluster NDB 7.0 releases due to a previous change in the
signalling format used between nodes. This issue was
corrected in MySQL Cluster NDB 7.0.9.
Once an NDB table had an
ALTER
ONLINE TABLE operation performed on it using a
MySQL Cluster NDB 6.3.x release, it could not be
upgraded online to MySQL Cluster NDB 7.0. This issue was
resolved in MySQL Cluster NDB 7.0.8. (See Bug#47542.)
Following an upgrade from MySQL Cluster NDB 6.3 to MySQL
Cluster NDB 7.0, if there were any tables having unique
indexes prior to the upgrade, attempts to create unique
indexes failed. This could also occur when performing
offline ALTER TABLE
operations on tables having indexes that were not
dropped as a result of the ALTER
TABLE . This issue was due to a change in the
way that NDB tracked unique
indexes internally, and was resolved in MySQL Cluster
NDB 7.0.9. (Bug#48416)
Workaround.
For upgrades to MySQL Cluster NDB 7.0 releases prior
to version 7.0.9, a workaround are available:
Following the upgrade, perform a second rolling
restart of the cluster before before performing any
ALTER TABLE operations
involving indexes.
A table created in a previous version of MySQL Cluster
does not automatically support
NDB -native default values
after the cluster is upgraded to MySQL Cluster NDB 7.1.0
or later. Such a table continues to use default values
supplied by the MySQL server until it is upgraded by
performing an offline ALTER
TABLE on it.
When upgrading to a MySQL Cluster NDB 7.1 or later
release from a MySQL Cluster NDB 7.0 release, you should
not attempt to create any new tables until all data
nodes are using the new ndbd or
ndbmtd binary. This is because the
older binaries do not provide support for native default
values; tables created with native default value support
cannot be used with NDB 7.0.x or earlier versions of the
software.
Due to an issue discovered after the release of MySQL
Cluster NDB 7.0.10 (Bug#50433), it is not possible to
perform an online upgrade from MySQL Cluster NDB 7.0.9b
and earlier MySQL Cluster NDB 7.0 releases to MySQL
Cluster NDB 7.0.10. Instead, you should upgrade your
MySQL Cluster NDB 7.0 cluster directly to MySQL Cluster
NDB 7.0.11 or later.
This issue did not appear to affect MySQL Cluster NDB
6.3, and it should be possible to upgrade online from
MySQL Cluster NDB 6.3 to MySQL Cluster NDB 7.0.10
without any problems other than those noted preciously.
17.3. MySQL Cluster Configuration
A MySQL server that is part of a MySQL Cluster differs in one chief
respect from a normal (nonclustered) MySQL server, in that it
employs the NDBCLUSTER storage engine.
This engine is also referred to simply as
NDB , and the two forms of the name are
synonymous.
To avoid unnecessary allocation of resources, the server is
configured by default with the NDB
storage engine disabled. To enable NDB ,
you must modify the server's my.cnf
configuration file, or start the server with the
--ndbcluster option.
For more information about
--ndbcluster and other MySQL server
options specific to MySQL Cluster, see
Section 17.3.4.2, “mysqld Command Options for MySQL Cluster”.
The MySQL server is a part of the cluster, so it also must know how
to access an MGM node to obtain the cluster configuration data. The
default behavior is to look for the MGM node on
localhost . However, should you need to specify
that its location is elsewhere, this can be done in
my.cnf or on the MySQL server command line.
Before the NDB storage engine can be
used, at least one MGM node must be operational, as well as any
desired data nodes.
NDB , the MySQL Cluster storage engine,
is available in binary distributions for Linux, Mac OS X, and
Solaris. We are working to support MySQL Cluster on all operating
systems supported by MySQL, including Windows. For information about
installing MySQL Cluster, see
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”.
17.3.1. Quick Test Setup of MySQL Cluster
To familiarize you with the basics, we will describe the simplest
possible configuration for a functional MySQL Cluster. After this,
you should be able to design your desired setup from the
information provided in the other relevant sections of this
chapter.
First, you need to create a configuration directory such as
/var/lib/mysql-cluster , by executing the
following command as the system root user:
shell> mkdir /var/lib/mysql-cluster
In this directory, create a file named
config.ini that contains the following
information. Substitute appropriate values for
HostName and DataDir as
necessary for your system.
# file "config.ini" - showing minimal setup consisting of 1 data node,
# 1 management server, and 3 MySQL servers.
# The empty default sections are not required, and are shown only for
# the sake of completeness.
# Data nodes must provide a hostname but MySQL Servers are not required
# to do so.
# If you don't know the hostname for your machine, use localhost.
# The DataDir parameter also has a default value, but it is recommended to
# set it explicitly.
# Note: [db], [api], and [mgm] are aliases for [ndbd], [mysqld], and [ndb_mgmd],
# respectively. [db] is deprecated and should not be used in new installations.
[ndbd default]
NoOfReplicas= 1
[mysqld default]
[ndb_mgmd default]
[tcp default]
[ndb_mgmd]
HostName= myhost.example.com
[ndbd]
HostName= myhost.example.com
DataDir= /var/lib/mysql-cluster
[mysqld]
[mysqld]
[mysqld]
You can now start the ndb_mgmd management
server. By default, it attempts to read the
config.ini file in its current working
directory, so change location into the directory where the file is
located and then invoke ndb_mgmd:
shell> cd /var/lib/mysql-cluster
shell> ndb_mgmd
Then start a single data node by running ndbd:
shell> ndbd
For command-line options which can be used when starting
ndbd, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
By default, ndbd looks for the management
server at localhost on port 1186.
Note
If you have installed MySQL from a binary tarball, you will need
to specify the path of the ndb_mgmd and
ndbd servers explicitly. (Normally, these
will be found in /usr/local/mysql/bin .)
Finally, change location to the MySQL data directory (usually
/var/lib/mysql or
/usr/local/mysql/data ), and make sure that
the my.cnf file contains the option necessary
to enable the NDB storage engine:
[mysqld]
ndbcluster
You can now start the MySQL server as usual:
shell> mysqld_safe --user=mysql &
Wait a moment to make sure the MySQL server is running properly.
If you see the notice mysql ended , check the
server's .err file to find out what went
wrong.
If all has gone well so far, you now can start using the cluster.
Connect to the server and verify that the
NDBCLUSTER storage engine is enabled:
shell> mysql
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 1 to server version: 5.1.45
Type 'help;' or '\h' for help. Type '\c' to clear the buffer.
mysql> SHOW ENGINES\G
...
*************************** 12. row ***************************
Engine: NDBCLUSTER
Support: YES
Comment: Clustered, fault-tolerant, memory-based tables
*************************** 13. row ***************************
Engine: NDB
Support: YES
Comment: Alias for NDBCLUSTER
...
The row numbers shown in the preceding example output may be
different from those shown on your system, depending upon how your
server is configured.
Try to create an NDBCLUSTER table:
shell> mysql
mysql> USE test;
Database changed
mysql> CREATE TABLE ctest (i INT) ENGINE=NDBCLUSTER;
Query OK, 0 rows affected (0.09 sec)
mysql> SHOW CREATE TABLE ctest \G
*************************** 1. row ***************************
Table: ctest
Create Table: CREATE TABLE `ctest` (
`i` int(11) default NULL
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
1 row in set (0.00 sec)
To check that your nodes were set up properly, start the
management client:
shell> ndb_mgm
Use the SHOW command from within the management
client to obtain a report on the cluster's status:
ndb_mgm> SHOW
Cluster Configuration
---------------------
[ndbd(NDB)] 1 node(s)
id=2 @127.0.0.1 (Version: 3.5.3, Nodegroup: 0, Master)
[ndb_mgmd(MGM)] 1 node(s)
id=1 @127.0.0.1 (Version: 3.5.3)
[mysqld(API)] 3 node(s)
id=3 @127.0.0.1 (Version: 3.5.3)
id=4 (not connected, accepting connect from any host)
id=5 (not connected, accepting connect from any host)
At this point, you have successfully set up a working MySQL
Cluster. You can now store data in the cluster by using any table
created with ENGINE=NDBCLUSTER or its alias
ENGINE=NDB .
17.3.2. MySQL Cluster Configuration Files
Configuring MySQL Cluster requires working with two files:
my.cnf : Specifies options for all MySQL
Cluster executables. This file, with which you should be
familiar with from previous work with MySQL, must be
accessible by each executable running in the cluster.
config.ini : This file, sometimes known as
the global configuration file, is read
only by the MySQL Cluster management server, which then
distributes the information contained therein to all processes
participating in the cluster. config.ini
contains a description of each node involved in the cluster.
This includes configuration parameters for data nodes and
configuration parameters for connections between all nodes in
the cluster. For a quick reference to the sections that can
appear in this file, and what sorts of configuration
parameters may be placed in each section, see
Sections of
the config.ini File.
Caching of configuration data.
Beginning with MySQL Cluster NDB 6.4.0, MySQL Cluster uses
stateful configuration. The global
configuration file is no longer read every time the management
server is restarted. Instead, the management server caches the
configuration the first time it is started, and thereafter, the
global confiuration file is read only when one of the following
items is true:
The management server is started using --initial option.
In this case, the global configuration file is re-read,
any existing cache files are deleted, and the management
server creates a new configuration cache.
The management server is started using --reload option.
In this case, the management server compares its cache
with the global configuration file. If they differ, the
management server creates a new configuration cache; any
existing configuration cache is preserved, but not used.
If the management server's cache and the global
configuration file contain the same configuration data,
then the existing cache is used, and no new cache is
created.
No configuration cache is found.
In this case, the management server reads the global
configuration file and creates a cache containing the
same configuration data as found in the file.
Configuration cache files.
Beginning with MySQL Cluster 6.4.0, the management server by
default creates configuration cache files in a directory named
mysql-cluster in the MySQL installation
directory. (If you build MySQL Cluster from source on a Unix
system, the default location is
/usr/local/mysql-cluster .) This can be
overridden at run time by starting the management server with
the --configdir option. Configuration cache
files are binary files named according to the pattern
ndb_node_id _config.bin.seq_id ,
where node_id is the management
server's node ID in the cluster, and
seq_id is a cache idenitifer. Cache
files are numbered sequentially using
seq_id , in the order in which they
are created. The management server uses the latest cache file as
determined by the seq_id .
Note
It is possible to roll back to a previous configuration by
deleting later configuration cache files, or by renaming an
earlier cache file so that it has a higher
seq_id . However, since configuration
cache files are written in a binary format, you should not
attempt to edit their contents by hand.
For more information about the --configdir ,
--initial , and --reload options
for the MySQL Cluster management server, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
We are continuously making improvements in Cluster configuration
and attempting to simplify this process. Although we strive to
maintain backward compatibility, there may be times when introduce
an incompatible change. In such cases we will try to let Cluster
users know in advance if a change is not backward compatible. If
you find such a change and we have not documented it, please
report it in the MySQL bugs database using the instructions given
in Section 1.7, “How to Report Bugs or Problems”.
17.3.2.1. MySQL Cluster Configuration — Basic Example
To support MySQL Cluster, you will need to update
my.cnf as shown in the following example.
You may also specify these parameters on the command line when
invoking the executables.
Note
The options shown here should not be confused with those that
are used in config.ini global
configuration files. Global configuration options are
discussed later in this section.
# my.cnf
# example additions to my.cnf for MySQL Cluster
# (valid in MySQL 5.1)
# enable ndbcluster storage engine, and provide connectstring for
# management server host (default port is 1186)
[mysqld]
ndbcluster
ndb-connectstring=ndb_mgmd.mysql.com
# provide connectstring for management server host (default port: 1186)
[ndbd]
connect-string=ndb_mgmd.mysql.com
# provide connectstring for management server host (default port: 1186)
[ndb_mgm]
connect-string=ndb_mgmd.mysql.com
# provide location of cluster configuration file
[ndb_mgmd]
config-file=/etc/config.ini
(For more information on connectstrings, see
Section 17.3.2.3, “The MySQL Cluster Connectstring”.)
# my.cnf
# example additions to my.cnf for MySQL Cluster
# (will work on all versions)
# enable ndbcluster storage engine, and provide connectstring for management
# server host to the default port 1186
[mysqld]
ndbcluster
ndb-connectstring=ndb_mgmd.mysql.com:1186
Important
Once you have started a mysqld process with
the NDBCLUSTER and
ndb-connectstring parameters in the
[mysqld] in the my.cnf
file as shown previously, you cannot execute any
CREATE TABLE or
ALTER TABLE statements without
having actually started the cluster. Otherwise, these
statements will fail with an error. This is by
design.
You may also use a separate [mysql_cluster]
section in the cluster my.cnf file for
settings to be read and used by all executables:
# cluster-specific settings
[mysql_cluster]
ndb-connectstring=ndb_mgmd.mysql.com:1186
For additional NDB variables that
can be set in the my.cnf file, see
Section 17.3.4.3, “MySQL Cluster System Variables”.
The MySQL Cluster global configuration file is named
config.ini by default. It is read by
ndb_mgmd at startup and can be placed
anywhere. Its location and name are specified by using
--config-file=path_name
on the ndb_mgmd command line. If the
configuration file is not specified, ndb_mgmd
by default tries to read a file named
config.ini located in the current working
directory.
The global configuration file for MySQL Cluster uses INI format,
which consists of sections preceded by section headings
(surrounded by square brackets), followed by the appropriate
parameter names and values. One deviation from the standard INI
format is that the parameter name and value can be separated by
a colon (“: ”) as well as the
equals sign (“= ”); however, the
equals sign is preferred. Another deviation is that sections are
not uniquely identified by section name. Instead, unique
sections (such as two different nodes of the same type) are
identified by a unique ID specified as a parameter within the
section.
Default values are defined for most parameters, and can also be
specified in config.ini . (Prior to MySQL
Cluster NDB 6.3.25 and MySQL Cluster NDB 7.0.6, there was no
default value for NoOfReplicas , which always
had to be specified explicitly in the [ndbd
default] section. Beginning with versions just stated,
the default value is 2, which is the recommended setting in most
common usage scenarios.) To create a default value section,
simply add the word default to the section
name. For example, an [ndbd] section contains
parameters that apply to a particular data node, whereas an
[ndbd default] section contains parameters
that apply to all data nodes. Suppose that all data nodes should
use the same data memory size. To configure them all, create an
[ndbd default] section that contains a
DataMemory line to specify the data memory
size.
The global configuration file must define the computers and
nodes involved in the cluster and on which computers these nodes
are located. An example of a simple configuration file for a
cluster consisting of one management server, two data nodes and
two MySQL servers is shown here:
# file "config.ini" - 2 data nodes and 2 SQL nodes
# This file is placed in the startup directory of ndb_mgmd (the
# management server)
# The first MySQL Server can be started from any host. The second
# can be started only on the host mysqld_5.mysql.com
[ndbd default]
NoOfReplicas= 2
DataDir= /var/lib/mysql-cluster
[ndb_mgmd]
Hostname= ndb_mgmd.mysql.com
DataDir= /var/lib/mysql-cluster
[ndbd]
HostName= ndbd_2.mysql.com
[ndbd]
HostName= ndbd_3.mysql.com
[mysqld]
[mysqld]
HostName= mysqld_5.mysql.com
Each node has its own section in the
config.ini file. For example, this cluster
has two data nodes, so the preceding configuration file contains
two [ndbd] sections defining these nodes.
Note
Do not place comments on the same line as a section heading in
the config.ini file; this causes the
management server not to start because it cannot parse the
configuration file in such cases.
Sections of the
config.ini File
There are six different sections that you can use in the
config.ini configuration file, as described
in the following list:
[computer] : Defines cluster hosts. This
is not required to configure a viable MySQL Cluster, but be
may used as a convenience when setting up a large cluster.
See Section 17.3.2.4, “Defining Computers in a MySQL Cluster”, for
more information.
[ndbd] : Defines a cluster data node
(ndbd process). See
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for
details.
[mysqld] : Defines the cluster's MySQL
server nodes (also called SQL or API nodes). For a
discussion of SQL node configuration, see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
[mgm] or [ndb_mgmd] :
Defines a cluster management server (MGM) node. For
information concerning the configuration of MGM nodes, see
Section 17.3.2.5, “Defining a MySQL Cluster Management Server”.
[tcp] : Defines a TCP/IP connection
between cluster nodes, with TCP/IP being the default
connection protocol. Normally, [tcp] or
[tcp default] sections are not required
to set up a MySQL Cluster, as the cluster handles this
automatically; however, it may be necessary in some
situations to override the defaults provided by the cluster.
See Section 17.3.2.8, “MySQL Cluster TCP/IP Connections”, for
information about available TCP/IP configuration parameters
and how to use them. (You may also find
Section 17.3.2.9, “MySQL Cluster TCP/IP Connections Using Direct Connections” to be
of interest in some cases.)
[shm] : Defines shared-memory connections
between nodes. In MySQL 5.1, it is enabled by
default, but should still be considered experimental. For a
discussion of SHM interconnects, see
Section 17.3.2.10, “MySQL Cluster Shared-Memory Connections”.
[sci] :Defines Scalable
Coherent Interface connections between cluster
data nodes. Such connections require software which, while
freely available, is not part of the MySQL Cluster
distribution, as well as specialized hardware. See
Section 17.3.2.11, “SCI Transport Connections in MySQL Cluster” for detailed
information about SCI interconnects.
You can define default values for each
section. All Cluster parameter names are case-insensitive, which
differs from parameters specified in my.cnf
or my.ini files.
17.3.2.2. Recommended Starting Configurations for MySQL Cluster NDB 6.2 and Later
Achieving the best performance from a MySQL Cluster depends on a
number of factors including the following:
MySQL Cluster software version
Numbers of data nodes and SQL nodes
Hardware
Operating system
Amount of data to be stored
Size and type of load under which the cluster is to operate
Therefore, obtaining an optimum configuration is likely to be an
iterative process, the outcome of which can vary widely with the
specifics of each MySQL Cluster deployment. Changes in
configuration are also likely to be indicated when changes are
made in the platform on which the cluster is run, or in
applications that use the MySQL Cluster's data. For these
reasons, it is not possible to offer a single configuration that
is ideal for all usage scenarios. However, in this section, we
provide recommended base configurations for MySQL Cluster NDB
6.2 and 6.3 that can serve as reasonable starting points.
Starting configuration for MySQL Cluster NDB 6.2.
The following is a recommended starting point for configuring
a cluster running MySQL Cluster NDB 6.2.
# TCP PARAMETERS
[tcp default]
SendBufferMemory=2M
ReceiveBufferMemory=2M
# Increasing the sizes of these 2 buffers beyond the default values
# helps prevent bottlenecks due to slow disk I/O.
# MANAGEMENT NODE PARAMETERS
[ndb_mgmd default]
DataDir=path/to/management/server/data/directory
# It is possible to use a different data directory for each management
# server, but for ease of administration it is preferable to be
# consistent.
[ndb_mgmd]
HostName=management-server-1-hostname
# Id=management-server-A-id
[ndb_mgmd]
HostName=management-server-2-hostname
# Using 2 management servers helps guarantee that there is always an
# arbitrator in the event of network partitioning, and so is
# recommended for high availability. Each management server must be
# identified by a HostName. You may for the sake of convenience specify
# a node ID for any management server, although one will be allocated
# for it automatically; if you do so, it must be in the range 1-255
# inclusive and must be unique among all IDs specified for cluster
# nodes.
# DATA NODE PARAMETERS
[ndbd default]
NoOfReplicas=2
# Using 2 replicas is recommended to guarantee availability of data;
# using only 1 replica does not provide any redundancy, which means
# that the failure of a single data node causes the entire cluster to
# shut down. We do not recommend using more than 2 replicas, since 2 is
# sufficient to provide high availability, and we do not currently test
# with greater values for this parameter.
LockPagesInMainMemory=1
# On Linux and Solaris systems, setting this parameter locks data node
# processes into memory. Doing so prevents them from swapping to disk,
# which can severely degrade cluster performance.
DataMemory=3072M
IndexMemory=384M
# The values provided for DataMemory and IndexMemory assume 4 GB RAM
# per data node. However, for best results, you should first calculate
# the memory that would be used based on the data you actually plan to
# store (you may find the ndb_size.pl utility helpful in estimating
# this), then allow an extra 20% over the calculated values. Naturally,
# you should ensure that each data node host has at least as much
# physical memory as the sum of these two values.
# ODirect=1
# Enabling this parameter causes NDBCLUSTER to try using O_DIRECT
# writes for local checkpoints and redo logs; this can reduce load on
# CPUs. We recommend doing so when using MySQL Cluster NDB 6.2.3 or
# newer on systems running Linux kernel 2.6 or later.
NoOfFragmentLogFiles=300
DataDir=path/to/data/node/data/directory
MaxNoOfConcurrentOperations=100000
TimeBetweenGlobalCheckpoints=1000
TimeBetweenEpochs=200
DiskCheckpointSpeed=10M
DiskCheckpointSpeedInRestart=100M
RedoBuffer=32M
# MaxNoOfLocalScans=64
MaxNoOfTables=1024
MaxNofOfOrderedIndexes=256
[ndbd]
HostName=data-node-A-hostname
# Id=data-node-A-id
[ndbd]
HostName=data-node-B-hostname
# Id=data-node-B-id
# You must have an [ndbd] section for every data node in the cluster;
# each of these sections must include a HostName. Each section may
# optionally include an Id for convenience, but in most cases, it is
# sufficient to allow the cluster to allocate node IDs dynamically. If
# you do specify the node ID for a data node, it must be in the range 1
# to 48 inclusive and must be unique among all IDs specified for
# cluster nodes.
# SQL NODE / API NODE PARAMETERS
[mysqld]
# HostName=SQL-node-1-hostname
# Id=sql-node-A-id
[mysqld]
[mysqld]
# Each API or SQL node that connects to the cluster requires a [mysqld]
# or [api] section of its own. Each such section defines a connection
# “slot”; you should have at least as many of these sections in the
# config.ini file as the total number of API nodes and SQL nodes that
# you wish to have connected to the cluster at any given time. There is
# no performance or other penalty for having extra slots available in
# case you find later that you want or need more API or SQL nodes to
# connect to the cluster at the same time.
# If no HostName is specified for a given [mysqld] or [api] section,
# then any API or SQL node may use that slot to connect to the
# cluster. You may wish to use an explicit HostName for one connection slot
# to guarantee that an API or SQL node from that host can always
# connect to the cluster. If you wish to prevent API or SQL nodes from
# connecting from other than a desired host or hosts, then use a
# HostName for every [mysqld] or [api] section in the config.ini file.
# You can if you wish define a node ID (Id parameter) for any API or
# SQL node, but this is not necessary; if you do so, it must be in the
# range 1 to 255 inclusive and must be unique among all IDs specified
# for cluster nodes.
Starting configuration for MySQL Cluster NDB 6.3.
The following is a recommended starting point for configuring
a cluster running MySQL Cluster NDB 6.3. It is similar to the
recommendation for MySQL Cluster NDB 6.2, with the addition of
parameters for better control of
NDBCLUSTER process threads.
# TCP PARAMETERS
[tcp default]
SendBufferMemory=2M
ReceiveBufferMemory=2M
# Increasing the sizes of these 2 buffers beyond the default values
# helps prevent bottlenecks due to slow disk I/O.
# MANAGEMENT NODE PARAMETERS
[ndb_mgmd default]
DataDir=path/to/management/server/data/directory
# It is possible to use a different data directory for each management
# server, but for ease of administration it is preferable to be
# consistent.
[ndb_mgmd]
HostName=management-server-1-hostname
# Id=management-server-A-id
[ndb_mgmd]
HostName=management-server-2-hostname
# Using 2 management servers helps guarantee that there is always an
# arbitrator in the event of network partitioning, and so is
# recommended for high availability. Each management server must be
# identified by a HostName. You may for the sake of convenience specify
# a node ID for any management server, although one will be allocated
# for it automatically; if you do so, it must be in the range 1-255
# inclusive and must be unique among all IDs specified for cluster
# nodes.
# DATA NODE PARAMETERS
[ndbd default]
NoOfReplicas=2
# Using 2 replicas is recommended to guarantee availability of data;
# using only 1 replica does not provide any redundancy, which means
# that the failure of a single data node causes the entire cluster to
# shut down. We do not recommend using more than 2 replicas, since 2 is
# sufficient to provide high availability, and we do not currently test
# with greater values for this parameter.
LockPagesInMainMemory=1
# On Linux and Solaris systems, setting this parameter locks data node
# processes into memory. Doing so prevents them from swapping to disk,
# which can severely degrade cluster performance.
DataMemory=3072M
IndexMemory=384M
# The values provided for DataMemory and IndexMemory assume 4 GB RAM
# per data node. However, for best results, you should first calculate
# the memory that would be used based on the data you actually plan to
# store (you may find the ndb_size.pl utility helpful in estimating
# this), then allow an extra 20% over the calculated values. Naturally,
# you should ensure that each data node host has at least as much
# physical memory as the sum of these two values.
# ODirect=1
# Enabling this parameter causes NDBCLUSTER to try using O_DIRECT
# writes for local checkpoints and redo logs; this can reduce load on
# CPUs. We recommend doing so when using MySQL Cluster NDB 6.2.3 or
# newer on systems running Linux kernel 2.6 or later.
NoOfFragmentLogFiles=300
DataDir=path/to/data/node/data/directory
MaxNoOfConcurrentOperations=100000
SchedulerSpinTimer=400
SchedulerExecutionTimer=100
RealTimeScheduler=1
# Setting these parameters allows you to take advantage of real-time scheduling
# of NDBCLUSTER threads (introduced in MySQL Cluster NDB 6.3.4) to get higher
# throughput.
TimeBetweenGlobalCheckpoints=1000
TimeBetweenEpochs=200
DiskCheckpointSpeed=10M
DiskCheckpointSpeedInRestart=100M
RedoBuffer=32M
# CompressedLCP=1
# CompressedBackup=1
# Enabling CompressedLCP and CompressedBackup causes, respectively, local
checkpoint files and backup files to be compressed, which can result in a space
savings of up to 50% over noncompressed LCPs and backups.
# MaxNoOfLocalScans=64
MaxNoOfTables=1024
MaxNofOfOrderedIndexes=256
[ndbd]
HostName=data-node-A-hostname
# Id=data-node-A-id
LockExecuteThreadToCPU=1
LockMaintThreadsToCPU=0
# On systems with multiple CPUs, these parameters can be used to lock NDBCLUSTER
# threads to specific CPUs
[ndbd]
HostName=data-node-B-hostname
# Id=data-node-B-id
LockExecuteThreadToCPU=1
LockMaintThreadsToCPU=0
# You must have an [ndbd] section for every data node in the cluster;
# each of these sections must include a HostName. Each section may
# optionally include an Id for convenience, but in most cases, it is
# sufficient to allow the cluster to allocate node IDs dynamically. If
# you do specify the node ID for a data node, it must be in the range 1
# to 48 inclusive and must be unique among all IDs specified for
# cluster nodes.
# SQL NODE / API NODE PARAMETERS
[mysqld]
# HostName=SQL-node-1-hostname
# Id=sql-node-A-id
[mysqld]
[mysqld]
# Each API or SQL node that connects to the cluster requires a [mysqld]
# or [api] section of its own. Each such section defines a connection
# “slot”; you should have at least as many of these sections in the
# config.ini file as the total number of API nodes and SQL nodes that
# you wish to have connected to the cluster at any given time. There is
# no performance or other penalty for having extra slots available in
# case you find later that you want or need more API or SQL nodes to
# connect to the cluster at the same time.
# If no HostName is specified for a given [mysqld] or [api] section,
# then any API or SQL node may use that slot to connect to the
# cluster. You may wish to use an explicit HostName for one connection slot
# to guarantee that an API or SQL node from that host can always
# connect to the cluster. If you wish to prevent API or SQL nodes from
# connecting from other than a desired host or hosts, then use a
# HostName for every [mysqld] or [api] section in the config.ini file.
# You can if you wish define a node ID (Id parameter) for any API or
# SQL node, but this is not necessary; if you do so, it must be in the
# range 1 to 255 inclusive and must be unique among all IDs specified
# for cluster nodes.
Recommended my.cnf options for SQL nodes.
MySQL Servers acting as MySQL Cluster SQL nodes must always be
started with the --ndbcluster
and --ndb-connectstring options, either on
the command line or in my.cnf . In
addition, set the following options for all
mysqld processes in the cluster, unless
your setup requires otherwise:
17.3.2.3. The MySQL Cluster Connectstring
With the exception of the MySQL Cluster management server
(ndb_mgmd), each node that is part of a MySQL
Cluster requires a connectstring that
points to the management server's location. This connectstring
is used in establishing a connection to the management server as
well as in performing other tasks depending on the node's role
in the cluster. The syntax for a connectstring is as follows:
[nodeid=node_id , ]host-definition [, host-definition [, ...]]
host-definition :
host_name [:port_number ]
node_id is an integer larger than 1 which
identifies a node in config.ini .
host_name is a string representing a
valid Internet host name or IP address.
port_number is an integer referring
to a TCP/IP port number.
example 1 (long): "nodeid=2,myhost1:1100,myhost2:1100,192.168.0.3:1200"
example 2 (short): "myhost1"
localhost:1186 is used as the default
connectstring value if none is provided. If
port_num is omitted from the
connectstring, the default port is 1186. This port should always
be available on the network because it has been assigned by IANA
for this purpose (see
http://www.iana.org/assignments/port-numbers for
details).
By listing multiple host definitions, it is possible to
designate several redundant management servers. A MySQL Cluster
data or API node attempts to contact successive management
servers on each host in the order specified, until a successful
connection has been established.
Beginning with MySQL Cluster NDB 6.3.19, it is also possible in
a connectstring to specify one or more bind addresses to be used
by nodes having multiple network interfaces for connecting to
management servers. A bind address consists of a hostname or
network address and an optional port number. This enhanced
syntax for connectstrings is shown here:
[nodeid=node_id , ]
[bind-address=host-definition , ]
host-definition [; bind-address=host-definition ]
host-definition [; bind-address=host-definition ]
[, ...]]
host-definition :
host_name [:port_number ]
If a single bind address is used in the connectstring
prior to specifying any management hosts,
then this address is used as the default for connecting to any
of them (unless overridden for a given management server; see
later in this section for an example). For example, the
following connectstring causes the node to use
192.168.178.242 regardless of the management
server to which it connects:
bind-address=192.168.178.242, poseidon:1186, perch:1186
If a bind address is specified following a
management host definition, then it is used only for connecting
to that management node. Consider the following connectstring:
poseidon:1186;bind-address=localhost, perch:1186;bind-address=192.168.178.242
In this case, the node uses localhost to
connect to the management server running on the host named
poseidon and
192.168.178.242 to connect to the management
server running on the host named perch .
You can specify a default bind address and then override this
default for one or more specific management hosts. In the
following example, localhost is used for
connecting to the management server running on host
poseidon ; since
192.168.178.242 is specified first (before
any management server definitions), it is the default bind
address and so is used for connecting to the management servers
on hosts perch and orca :
bind-address=192.168.178.242,poseidon:1186;bind-address=localhost,perch:1186,orca:2200
There are a number of different ways to specify the
connectstring:
Each executable has its own command-line option which
enables specifying the management server at startup. (See
the documentation for the respective executable.)
It is also possible to set the connectstring for all nodes
in the cluster at once by placing it in a
[mysql_cluster] section in the management
server's my.cnf file.
For backward compatibility, two other options are available,
using the same syntax:
Set the NDB_CONNECTSTRING environment
variable to contain the connectstring.
Write the connectstring for each executable into a text
file named Ndb.cfg and place this
file in the executable's startup directory.
However, these are now deprecated and should not be used for
new installations.
The recommended method for specifying the connectstring is to
set it on the command line or in the my.cnf
file for each executable.
The maximum length of a connectstring is 1024 characters.
17.3.2.4. Defining Computers in a MySQL Cluster
The [computer] section has no real
significance other than serving as a way to avoid the need of
defining host names for each node in the system. All parameters
mentioned here are required.
Id
This is a unique identifier, used to refer to the host
computer elsewhere in the configuration file.
Important
The computer ID is not the same as
the node ID used for a management, API, or data node.
Unlike the case with node IDs, you cannot use
NodeId in place of
Id in the [computer]
section of the config.ini file.
HostName
This is the computer's hostname or IP address.
17.3.2.5. Defining a MySQL Cluster Management Server
The [ndb_mgmd] section is used to configure
the behavior of the management server. [mgm]
can be used as an alias; the two section names are equivalent.
All parameters in the following list are optional and assume
their default values if omitted.
Note
If neither the ExecuteOnComputer nor the
HostName parameter is present, the default
value localhost will be assumed for both.
Id
Each node in the cluster has a unique identity. For a
management node, this is represented by an integer value in
the range 1 to 63 inclusive (previous to MySQL Cluster NDB
6.1.1), or in the range 1 to 255 inclusive (MySQL Cluster
NDB 6.1.1 and later). This ID is used by all internal
cluster messages for addressing the node, and so must be
unique for each MySQL Cluster node, regardless of the type
of node.
Note
Data node IDs must be less than 49, regardless of the
MySQL Cluster version used. If you plan to deploy a large
number of data nodes, it is a good idea to limit the node
IDs for management nodes (and API nodes) to values greater
than 48.
This parameter can also be written as
NodeId , although the short form is
sufficient (and preferred for this reason).
ExecuteOnComputer
This refers to the Id set for one of the
computers defined in a [computer] section
of the config.ini file.
PortNumber
This is the port number on which the management server
listens for configuration requests and management commands.
HostName
Specifying this parameter defines the hostname of the
computer on which the management node is to reside. To
specify a hostname other than localhost ,
either this parameter or
ExecuteOnComputer is required.
LogDestination
This parameter specifies where to send cluster logging
information. There are three options in this regard —
CONSOLE , SYSLOG , and
FILE — with FILE
being the default:
CONSOLE outputs the log to
stdout :
CONSOLE
SYSLOG sends the log to a
syslog facility, possible values
being one of auth ,
authpriv , cron ,
daemon , ftp ,
kern , lpr ,
mail , news ,
syslog , user ,
uucp , local0 ,
local1 , local2 ,
local3 , local4 ,
local5 , local6 , or
local7 .
Note
Not every facility is necessarily supported by every
operating system.
SYSLOG:facility=syslog
FILE pipes the cluster log output to
a regular file on the same machine. The following values
can be specified:
filename : The name of the log
file.
maxsize : The maximum size (in
bytes) to which the file can grow before logging
rolls over to a new file. When this occurs, the old
log file is renamed by appending
.N to the file name,
where N is the next
number not yet used with this name.
maxfiles : The maximum number of
log files.
FILE:filename=cluster.log,maxsize=1000000,maxfiles=6
The default value for the FILE
parameter is
FILE:filename=ndb_node_id _cluster.log,maxsize=1000000,maxfiles=6 ,
where node_id is the ID of
the node.
It is possible to specify multiple log destinations
separated by semicolons as shown here:
CONSOLE;SYSLOG:facility=local0;FILE:filename=/var/log/mgmd
ArbitrationRank
This parameter is used to define which nodes can act as
arbitrators. Only management nodes and SQL nodes can be
arbitrators. ArbitrationRank can take one
of the following values:
0 : The node will never be used as an
arbitrator.
1 : The node has high priority; that
is, it will be preferred as an arbitrator over
low-priority nodes.
2 : Indicates a low-priority node
which be used as an arbitrator only if a node with a
higher priority is not available for that purpose.
Normally, the management server should be configured as an
arbitrator by setting its ArbitrationRank
to 1 (the default for management nodes) and those for all
SQL nodes to 0 (the default for SQL nodes).
Beginning with MySQL 5.1.16 and MySQL Cluster NDB 6.1.3, it
is possible to disable arbitration completely by setting
ArbitrationRank to 0 on all management
and SQL nodes. In MySQL Cluster NDB 7.0.7 and later
releases, you can also control arbitration by overriding
this parameter; to do this, set the
Arbitration
parameter in the [ndbd default] section
of the config.ini global configuration
file.
ArbitrationDelay
An integer value which causes the management server's
responses to arbitration requests to be delayed by that
number of milliseconds. By default, this value is 0; it is
normally not necessary to change it.
DataDir
This specifies the directory where output files from the
management server will be placed. These files include
cluster log files, process output files, and the daemon's
process ID (PID) file. (For log files, this location can be
overridden by setting the FILE parameter
for LogDestination as discussed
previously in this section.)
The default value for this parameter is the directory in
which ndb_mgmd is located.
HeartbeatThreadPriority
Beginning with MySQL Cluster NDB 6.3.32, MySQL Cluster NDB
7.0.13, and MySQL Cluster NDB 7.1.2, it is possible to use
this parameter to set the scheduling policy and priority of
heartbeat threads for management and API nodes.
The syntax for setting this parameter is shown here:
HeartbeatThreadPriority = policy [, priority ]
policy :
{FIFO | RR}
When setting this parameter, you must specify a policy. This
is one of FIFO (first in, first in) or
RR (round robin). This followed
optionally by the priority (an integer).
TotalSendBufferMemory
This parameter is available beginning with MySQL Cluster NDB
6.4.0. It is used to determine the total amount of memory to
allocate on this node for shared send buffer memory among
all configured transporters.
If this parameter is set, its minimum allowed value is 256K;
the maxmimum is 4294967039. For more detailed information
about the behavior and use of
TotalSendBufferMemory and configuring
send buffer memory parameters in MySQL Cluster NDB 6.4.0 and
later, see
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”.
Note
After making changes in a management node's
configuration, it is necessary to perform a rolling restart of
the cluster in order for the new configuration to take effect.
To add new management servers to a running MySQL Cluster, it
is also necessary to perform a rolling restart of all cluster
nodes after modifying any existing
config.ini files. For more information
about issues arising when using multiple management nodes, see
Section 17.1.5.10, “Limitations Relating to Multiple MySQL Cluster Nodes”.
17.3.2.6. Defining MySQL Cluster Data Nodes
The [ndbd] and [ndbd
default] sections are used to configure the behavior
of the cluster's data nodes.
[ndbd] and [ndbd default]
are always used as the section names whether you are using
ndbd or (in MySQL Cluster NDB 6.4.0 and
later) ndbmtd binaries for the data node
processes.
There are many parameters which control buffer sizes, pool
sizes, timeouts, and so forth. The only mandatory parameters
are:
Either ExecuteOnComputer or
HostName , which must be defined in the
local [ndbd] section.
The parameter NoOfReplicas , which must be
defined in the[ndbd default] section, as
it is common to all Cluster data nodes.
Note
It is no longer strictly necessary to set
NoOfReplicas starting with MySQL Cluster
NDB 6.3.25 and MySQL Cluster NDB 7.0.6, where it acquires a
default value (2). However, it remains good practice to set it
explicitly.
Most data node parameters are set in the [ndbd
default] section. Only those parameters explicitly
stated as being able to set local values are allowed to be
changed in the [ndbd] section. Where present,
HostName , Id and
ExecuteOnComputer must
be defined in the local [ndbd] section, and
not in any other section of config.ini . In
other words, settings for these parameters are specific to one
data node.
For those parameters affecting memory usage or buffer sizes, it
is possible to use K , M ,
or G as a suffix to indicate units of 1024,
1024?1024, or 1024?1024?1024. (For example,
100K means 100 ? 1024 = 102400.)
Parameter names and values are currently case-sensitive.
Information about configuration parameters specific to MySQL
Cluster Disk Data tables can be found later in this section.
Beginning with MySQL Cluster NDB 6.4.0, all of these parameters
also apply to ndbmtd (the multi-threaded
version of ndbd). An additional data node
configuration parameter
MaxNoOfExecutionThreads applies to
ndbmtd only, and has no effect when used with
ndbd. For more information, see
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”.
Identifying data nodes.
The Id value (that is, the data node
identifier) can be allocated on the command line when the node
is started or in the configuration file.
Id
This is the node ID used as the address of the node for all
cluster internal messages. For data nodes, this is an
integer in the range 1 to 48 inclusive. Each node in the
cluster must have a unique identifier.
This parameter can also be written as
NodeId , although the short form is
sufficient (and preferred for this reason).
ExecuteOnComputer
This refers to the Id set for one of the
computers defined in a [computer]
section.
HostName
Specifying this parameter defines the hostname of the
computer on which the data node is to reside. To specify a
hostname other than localhost , either
this parameter or ExecuteOnComputer is
required.
ServerPort
Each node in the cluster uses a port to connect to other
nodes. By default, this port is allocated dynamically in
such a way as to ensure that no two nodes on the same host
computer receive the same port number, so it should normally
not be necessary to specify a value for this parameter.
However, if you need to be able to open specific ports in a
firewall to permit communication between data nodes and API
nodes (including SQL nodes), you can set this parameter to
the number of the desired port in an
[ndbd] section or (if you need to do this
for multiple data nodes) the [ndbd
default] section of the
config.ini file, and then open the port
having that number for incoming connections from SQL nodes,
API nodes, or both.
TcpBind_INADDR_ANY
Setting this parameter to TRUE or
1 binds IP_ADDR_ANY so
that connections can be made from anywhere (for
autogenerated connections). The default is
FALSE (0 ).
This parameter was added in MySQL Cluster NDB 6.2.0.
NodeGroup
This parameter can be used to assign a data node to a
specific node group. It is read only when the cluster is
started for the first time, and cannot be used to reassign a
data node to a different node group online. It is generally
not desirable to use this parameter in the [ndbd
default] section of the
config.ini file, and care must be taken
not to assign nodes to node groups in such a way that an
invalid numbers of nodes are assigned to any node groups.
The NodeGroup parameter is chiefly
intended for use in adding a new node group to a running
MySQL Cluster without having to perform a rolling restart.
For this purpose, you should set it to 65535 (the maximum
value). You are not required to set a
NodeGroup value for all cluster data
nodes, only for those nodes which are to be started and
added to the cluster as a new node group at a later time.
For more information, see
Section 17.5.11.3, “Adding MySQL Cluster Data Nodes Online: Detailed Example”.
This parameter was added in MySQL Cluster NDB 6.4.0.
NoOfReplicas
This global parameter can be set only in the [ndbd
default] section, and defines the number of
replicas for each table stored in the cluster. This
parameter also specifies the size of node groups. A node
group is a set of nodes all storing the same information.
Node groups are formed implicitly. The first node group is
formed by the set of data nodes with the lowest node IDs,
the next node group by the set of the next lowest node
identities, and so on. By way of example, assume that we
have 4 data nodes and that NoOfReplicas
is set to 2. The four data nodes have node IDs 2, 3, 4 and
5. Then the first node group is formed from nodes 2 and 3,
and the second node group by nodes 4 and 5. It is important
to configure the cluster in such a manner that nodes in the
same node groups are not placed on the same computer because
a single hardware failure would cause the entire cluster to
fail.
If no node IDs are provided, the order of the data nodes
will be the determining factor for the node group. Whether
or not explicit assignments are made, they can be viewed in
the output of the management client's
SHOW command.
Prior to MySQL Cluster NDB 6.3.25 and MySQL Cluster NDB
7.0.6, there was no default value for
NoOfReplicas ; beginning with these
versions, the default value is 2, which is the recommended
setting in most common usage scenarios. (Bug#44746)
The maximum possible value is 4; currently, only
the values 1 and 2 are actually supported (see
Bug#18621).
Important
Setting NoOfReplicas to 1 means that
there is only a single copy of all Cluster data; in this
case, the loss of a single data node causes the cluster to
fail because there are no additional copies of the data
stored by that node.
The value for this parameter must divide evenly into the
number of data nodes in the cluster. For example, if there
are two data nodes, then NoOfReplicas
must be equal to either 1 or 2, since 2/3 and 2/4 both yield
fractional values; if there are four data nodes, then
NoOfReplicas must be equal to 1, 2, or 4.
DataDir
This parameter specifies the directory where trace files,
log files, pid files and error logs are placed.
The default is the data node process working directory.
FileSystemPath
This parameter specifies the directory where all files
created for metadata, REDO logs, UNDO logs (for Disk Data
tables), and data files are placed. The default is the
directory specified by DataDir .
Note
This directory must exist before the
ndbd process is initiated.
The recommended directory hierarchy for MySQL Cluster
includes /var/lib/mysql-cluster , under
which a directory for the node's file system is created. The
name of this subdirectory contains the node ID. For example,
if the node ID is 2, this subdirectory is named
ndb_2_fs .
BackupDataDir
This parameter specifies the directory in which backups are
placed. If omitted, the default backup location is the
directory named BACKUP under the
location specified by the FileSystemPath
parameter. (See above.)
Data Memory, Index Memory, and String
Memory
DataMemory and IndexMemory
are [ndbd] parameters specifying the size of
memory segments used to store the actual records and their
indexes. In setting values for these, it is important to
understand how DataMemory and
IndexMemory are used, as they usually need to
be updated to reflect actual usage by the cluster:
DataMemory
This parameter defines the amount of space (in bytes)
available for storing database records. The entire amount
specified by this value is allocated in memory, so it is
extremely important that the machine has sufficient physical
memory to accommodate it.
The memory allocated by DataMemory is
used to store both the actual records and indexes. There is
a 16-byte overhead on each record; an additional amount for
each record is incurred because it is stored in a 32KB page
with 128 byte page overhead (see below). There is also a
small amount wasted per page due to the fact that each
record is stored in only one page.
For variable-size table attributes in MySQL 5.1, the data is
stored on separate datapages, allocated from
DataMemory . Variable-length records use a
fixed-size part with an extra overhead of 4 bytes to
reference the variable-size part. The variable-size part has
2 bytes overhead plus 2 bytes per attribute.
The maximum record size is currently 8052 bytes.
The memory space defined by DataMemory is
also used to store ordered indexes, which use about 10 bytes
per record. Each table row is represented in the ordered
index. A common error among users is to assume that all
indexes are stored in the memory allocated by
IndexMemory , but this is not the case:
Only primary key and unique hash indexes use this memory;
ordered indexes use the memory allocated by
DataMemory . However, creating a primary
key or unique hash index also creates an ordered index on
the same keys, unless you specify USING
HASH in the index creation statement. This can be
verified by running ndb_desc -d
db_name
table_name in the
management client.
The memory space allocated by DataMemory
consists of 32KB pages, which are allocated to table
fragments. Each table is normally partitioned into the same
number of fragments as there are data nodes in the cluster.
Thus, for each node, there are the same number of fragments
as are set in NoOfReplicas .
In addition, due to the way in which new pages are allocated
when the capacity of the current page is exhausted, there is
an additional overhead of approximately 18.75%. When more
DataMemory is required, more than one new
page is allocated, according to the following formula:
number of new pages = FLOOR(number of current pages ? 0.1875) + 1
For example, if 15 pages are currently allocated to a given
table and an insert to this table requires additional
storage space, the number of new pages allocated to the
table is FLOOR(15 ? 0.1875) + 1 =
FLOOR(2.8125) + 1 = 2 + 1 =
3 . Now 15 + 3 = 18 memory pages are
allocated to the table. When the last of these 18 pages
becomes full, FLOOR(18 ? 0.1875) + 1
= FLOOR(3.3750) + 1 = 3 + 1 =
4 new pages are allocated, so the total number of
pages allocated to the table is now 22.
Note
The “18.75% + 1” overhead is no longer
required beginning with MySQL Cluster NDB 6.2.3 and MySQL
Cluster NDB 6.3.0.
Once a page has been allocated, it is currently not possible
to return it to the pool of free pages, except by deleting
the table. (This also means that
DataMemory pages, once allocated to a
given table, cannot be used by other tables.) Performing a
node recovery also compresses the partition because all
records are inserted into empty partitions from other live
nodes.
The DataMemory memory space also contains
UNDO information: For each update, a copy of the unaltered
record is allocated in the DataMemory .
There is also a reference to each copy in the ordered table
indexes. Unique hash indexes are updated only when the
unique index columns are updated, in which case a new entry
in the index table is inserted and the old entry is deleted
upon commit. For this reason, it is also necessary to
allocate enough memory to handle the largest transactions
performed by applications using the cluster. In any case,
performing a few large transactions holds no advantage over
using many smaller ones, for the following reasons:
Large transactions are not any faster than smaller ones
Large transactions increase the number of operations
that are lost and must be repeated in event of
transaction failure
Large transactions use more memory
The default value for DataMemory is 80MB;
the minimum is 1MB. There is no maximum size, but in reality
the maximum size has to be adapted so that the process does
not start swapping when the limit is reached. This limit is
determined by the amount of physical RAM available on the
machine and by the amount of memory that the operating
system may commit to any one process. 32-bit operating
systems are generally limited to 2–4GB per process;
64-bit operating systems can use more. For large databases,
it may be preferable to use a 64-bit operating system for
this reason.
IndexMemory
This parameter controls the amount of storage used for hash
indexes in MySQL Cluster. Hash indexes are always used for
primary key indexes, unique indexes, and unique constraints.
Note that when defining a primary key and a unique index,
two indexes will be created, one of which is a hash index
used for all tuple accesses as well as lock handling. It is
also used to enforce unique constraints.
The size of the hash index is 25 bytes per record, plus the
size of the primary key. For primary keys larger than 32
bytes another 8 bytes is added.
The default value for IndexMemory is
18MB. The minimum is 1MB.
StringMemory
This parameter determines how much memory is allocated for
strings such as table names, and is specified in an
[ndbd] or [ndbd
default] section of the
config.ini file. A value between
0 and 100 inclusive is
interpreted as a percent of the maximum default value, which
is calculated based on a number of factors including the
number of tables, maximum table name size, maximum size of
.FRM files,
MaxNoOfTriggers , maximum column name
size, and maximum default column value. In general it is
safe to assume that the maximum default value is
approximately 5 MB for a MySQL Cluster having 1000 tables.
A value greater than 100 is interpreted
as a number of bytes.
Beginning with MySQL Cluster NDB 6.2.18, MySQL Cluster NDB
6.3.24, and MySQL Cluster NDB 7.0.5, the default value is 25
— that is, 25 percent of the default maximum, or
approximately 25 KB. (Previously, the default value was 5
beginning with MySQL 5.1.6; prior to MySQL 5.1.6, the
default was 0.)
Under most circumstances, the default value should be
sufficient, but when you have a great many Cluster tables
(1000 or more), it is possible to get Error 773
Out of string memory, please modify StringMemory
config parameter: Permanent error: Schema error,
in which case you should increase this value.
25 (25 percent) is not excessive, and
should prevent this error from recurring in all but the most
extreme conditions.
The following example illustrates how memory is used for a
table. Consider this table definition:
CREATE TABLE example (
a INT NOT NULL,
b INT NOT NULL,
c INT NOT NULL,
PRIMARY KEY(a),
UNIQUE(b)
) ENGINE=NDBCLUSTER;
For each record, there are 12 bytes of data plus 12 bytes
overhead. Having no nullable columns saves 4 bytes of overhead.
In addition, we have two ordered indexes on columns
a and b consuming roughly
10 bytes each per record. There is a primary key hash index on
the base table using roughly 29 bytes per record. The unique
constraint is implemented by a separate table with
b as primary key and a as
a column. This other table consumes an additional 29 bytes of
index memory per record in the example table
as well 8 bytes of record data plus 12 bytes of overhead.
Thus, for one million records, we need 58MB for index memory to
handle the hash indexes for the primary key and the unique
constraint. We also need 64MB for the records of the base table
and the unique index table, plus the two ordered index tables.
You can see that hash indexes takes up a fair amount of memory
space; however, they provide very fast access to the data in
return. They are also used in MySQL Cluster to handle uniqueness
constraints.
Currently, the only partitioning algorithm is hashing and
ordered indexes are local to each node. Thus, ordered indexes
cannot be used to handle uniqueness constraints in the general
case.
An important point for both IndexMemory and
DataMemory is that the total database size is
the sum of all data memory and all index memory for each node
group. Each node group is used to store replicated information,
so if there are four nodes with two replicas, there will be two
node groups. Thus, the total data memory available is 2 ?
DataMemory for each data node.
It is highly recommended that DataMemory and
IndexMemory be set to the same values for all
nodes. Data distribution is even over all nodes in the cluster,
so the maximum amount of space available for any node can be no
greater than that of the smallest node in the cluster.
DataMemory and IndexMemory
can be changed, but decreasing either of these can be risky;
doing so can easily lead to a node or even an entire MySQL
Cluster that is unable to restart due to there being
insufficient memory space. Increasing these values should be
acceptable, but it is recommended that such upgrades are
performed in the same manner as a software upgrade, beginning
with an update of the configuration file, and then restarting
the management server followed by restarting each data node in
turn.
Updates do not increase the amount of index memory used. Inserts
take effect immediately; however, rows are not actually deleted
until the transaction is committed.
Transaction parameters.
The next three [ndbd] parameters that we
discuss are important because they affect the number of
parallel transactions and the sizes of transactions that can
be handled by the system.
MaxNoOfConcurrentTransactions sets the
number of parallel transactions possible in a node.
MaxNoOfConcurrentOperations sets the number
of records that can be in update phase or locked
simultaneously.
Both of these parameters (especially
MaxNoOfConcurrentOperations ) are likely
targets for users setting specific values and not using the
default value. The default value is set for systems using small
transactions, to ensure that these do not use excessive memory.
MaxNoOfConcurrentTransactions
Each cluster data node requires a transaction record for
each active transaction in the cluster. The task of
coordinating transactions is distributed among all of the
data nodes. The total number of transaction records in the
cluster is the number of transactions in any given node
times the number of nodes in the cluster.
Transaction records are allocated to individual MySQL
servers. Each connection to a MySQL server requires at least
one transaction record, plus an additional transaction
object per table accessed by that connection. This means
that a reasonable minimum for this parameter is
MaxNoOfConcurrentTransactions =
(maximum number of tables accessed in any single transaction + 1)
* number of cluster SQL nodes
Suppose that there are 4 SQL nodes using the cluster. A
single join involving 5 tables requires 6 transaction
records; if there are 5 such joins in a transaction, then 5
* 6 = 30 transaction records are required for this
transaction, per MySQL server, or 30 * 4 = 120 transaction
records total.
This parameter must be set to the same value for all cluster
data nodes. This is due to the fact that, when a data node
fails, the oldest surviving node re-creates the transaction
state of all transactions that were ongoing in the failed
node.
Changing the value of
MaxNoOfConcurrentTransactions requires a
complete shutdown and restart of the cluster.
The default value is 4096.
MaxNoOfConcurrentOperations
It is a good idea to adjust the value of this parameter
according to the size and number of transactions. When
performing transactions of only a few operations each and
not involving a great many records, there is no need to set
this parameter very high. When performing large transactions
involving many records need to set this parameter higher.
Records are kept for each transaction updating cluster data,
both in the transaction coordinator and in the nodes where
the actual updates are performed. These records contain
state information needed to find UNDO records for rollback,
lock queues, and other purposes.
This parameter should be set to the number of records to be
updated simultaneously in transactions, divided by the
number of cluster data nodes. For example, in a cluster
which has four data nodes and which is expected to handle
1,000,000 concurrent updates using transactions, you should
set this value to 1000000 / 4 = 250000.
Read queries which set locks also cause operation records to
be created. Some extra space is allocated within individual
nodes to accommodate cases where the distribution is not
perfect over the nodes.
When queries make use of the unique hash index, there are
actually two operation records used per record in the
transaction. The first record represents the read in the
index table and the second handles the operation on the base
table.
The default value is 32768.
This parameter actually handles two values that can be
configured separately. The first of these specifies how many
operation records are to be placed with the transaction
coordinator. The second part specifies how many operation
records are to be local to the database.
A very large transaction performed on an eight-node cluster
requires as many operation records in the transaction
coordinator as there are reads, updates, and deletes
involved in the transaction. However, the operation records
of the are spread over all eight nodes. Thus, if it is
necessary to configure the system for one very large
transaction, it is a good idea to configure the two parts
separately. MaxNoOfConcurrentOperations
will always be used to calculate the number of operation
records in the transaction coordinator portion of the node.
It is also important to have an idea of the memory
requirements for operation records. These consume about 1KB
per record.
MaxNoOfLocalOperations
By default, this parameter is calculated as 1.1 ?
MaxNoOfConcurrentOperations . This fits
systems with many simultaneous transactions, none of them
being very large. If there is a need to handle one very
large transaction at a time and there are many nodes, it is
a good idea to override the default value by explicitly
specifying this parameter.
Transaction temporary storage.
The next set of [ndbd] parameters is used
to determine temporary storage when executing a statement that
is part of a Cluster transaction. All records are released
when the statement is completed and the cluster is waiting for
the commit or rollback.
The default values for these parameters are adequate for most
situations. However, users with a need to support transactions
involving large numbers of rows or operations may need to
increase these values to enable better parallelism in the
system, whereas users whose applications require relatively
small transactions can decrease the values to save memory.
MaxNoOfConcurrentIndexOperations
For queries using a unique hash index, another temporary set
of operation records is used during a query's execution
phase. This parameter sets the size of that pool of records.
Thus, this record is allocated only while executing a part
of a query. As soon as this part has been executed, the
record is released. The state needed to handle aborts and
commits is handled by the normal operation records, where
the pool size is set by the parameter
MaxNoOfConcurrentOperations .
The default value of this parameter is 8192. Only in rare
cases of extremely high parallelism using unique hash
indexes should it be necessary to increase this value. Using
a smaller value is possible and can save memory if the DBA
is certain that a high degree of parallelism is not required
for the cluster.
MaxNoOfFiredTriggers
The default value of MaxNoOfFiredTriggers
is 4000, which is sufficient for most situations. In some
cases it can even be decreased if the DBA feels certain the
need for parallelism in the cluster is not high.
A record is created when an operation is performed that
affects a unique hash index. Inserting or deleting a record
in a table with unique hash indexes or updating a column
that is part of a unique hash index fires an insert or a
delete in the index table. The resulting record is used to
represent this index table operation while waiting for the
original operation that fired it to complete. This operation
is short-lived but can still require a large number of
records in its pool for situations with many parallel write
operations on a base table containing a set of unique hash
indexes.
TransactionBufferMemory
The memory affected by this parameter is used for tracking
operations fired when updating index tables and reading
unique indexes. This memory is used to store the key and
column information for these operations. It is only very
rarely that the value for this parameter needs to be altered
from the default.
The default value for
TransactionBufferMemory is 1MB.
Normal read and write operations use a similar buffer, whose
usage is even more short-lived. The compile-time parameter
ZATTRBUF_FILESIZE (found in
ndb/src/kernel/blocks/Dbtc/Dbtc.hpp )
set to 4000 ? 128 bytes (500KB). A similar buffer for
key information, ZDATABUF_FILESIZE (also
in Dbtc.hpp ) contains 4000 ? 16 =
62.5KB of buffer space. Dbtc is the
module that handles transaction coordination.
Scans and buffering.
There are additional [ndbd] parameters in
the Dblqh module (in
ndb/src/kernel/blocks/Dblqh/Dblqh.hpp )
that affect reads and updates. These include
ZATTRINBUF_FILESIZE , set by default to
10000 ? 128 bytes (1250KB) and
ZDATABUF_FILE_SIZE , set by default to
10000*16 bytes (roughly 156KB) of buffer space. To date, there
have been neither any reports from users nor any results from
our own extensive tests suggesting that either of these
compile-time limits should be increased.
MaxNoOfConcurrentScans
This parameter is used to control the number of parallel
scans that can be performed in the cluster. Each transaction
coordinator can handle the number of parallel scans defined
for this parameter. Each scan query is performed by scanning
all partitions in parallel. Each partition scan uses a scan
record in the node where the partition is located, the
number of records being the value of this parameter times
the number of nodes. The cluster should be able to sustain
MaxNoOfConcurrentScans scans concurrently
from all nodes in the cluster.
Scans are actually performed in two cases. The first of
these cases occurs when no hash or ordered indexes exists to
handle the query, in which case the query is executed by
performing a full table scan. The second case is encountered
when there is no hash index to support the query but there
is an ordered index. Using the ordered index means executing
a parallel range scan. The order is kept on the local
partitions only, so it is necessary to perform the index
scan on all partitions.
The default value of
MaxNoOfConcurrentScans is 256. The
maximum value is 500.
MaxNoOfLocalScans
Specifies the number of local scan records if many scans are
not fully parallelized. If the number of local scan records
is not provided, it is calculated as the product of
MaxNoOfConcurrentScans and the number of
data nodes in the system. The minimum value is 32.
BatchSizePerLocalScan
This parameter is used to calculate the number of lock
records used to handle concurrent scan operations.
The default value is 64; this value has a strong connection
to the ScanBatchSize defined in the SQL
nodes.
LongMessageBuffer
This is an internal buffer used for passing messages within
individual nodes and between nodes. Although it is highly
unlikely that this would need to be changed, it is
configurable. In MySQL Cluster NDB 6.4.3 and earlier, the
default is 1MB; beginning with MySQL Cluster NDB 7.0.4, it
is 4MB.
Memory Allocation
MaxAllocate
This is the maximum size of the memory unit to use when
allocating memory for tables. In cases where
NDB gives Out of
memory errors, but it is evident by examining the
cluster logs or the output of DUMP 1000 (see
DUMP 1000 ) that all
available memory has not yet been used, you can increase the
value of this parameter (or MaxNoOfTables , or
both) in order to cause NDB to make
sufficient memory available.
This parameter was introduced in MySQL 5.1.20, MySQL Cluster NDB
6.1.12 and MySQL Cluster NDB 6.2.3.
Logging and checkpointing
The following [ndbd] parameters control log
and checkpoint behavior.
NoOfFragmentLogFiles
This parameter sets the number of REDO log files for the
node, and thus the amount of space allocated to REDO
logging. Because the REDO log files are organized in a ring,
it is extremely important that the first and last log files
in the set (sometimes referred to as the “head”
and “tail” log files, respectively) do not
meet. When these approach one another too closely, the node
begins aborting all transactions encompassing updates due to
a lack of room for new log records.
A REDO log record is not removed until
the required number of local checkpoints has been completed
since that log record was inserted (prior to MySQL Cluster
NDB 6.3.8, this was 3 local checkpoints; in later versions
of MySQL Cluster, only 2 local checkpoints are necessary).
Checkpointing frequency is determined by its own set of
configuration parameters discussed elsewhere in this
chapter.
How these parameters interact and proposals for how to
configure them are discussed in
Section 17.3.2.12, “Configuring MySQL Cluster Parameters for Local Checkpoints”.
The default parameter value is 16, which by default means 16
sets of 4 16MB files for a total of 1024MB. Beginning with
MySQL Cluster NDB 6.1.1, the size of the individual log
files is configurable using the
FragmentLogFileSize parameter; more
information about this parameter can be found
here.
In scenarios requiring a great many updates, the value for
NoOfFragmentLogFiles may need to be set
as high as 300 or even higher to provide sufficient space
for REDO logs.
If the checkpointing is slow and there are so many writes to
the database that the log files are full and the log tail
cannot be cut without jeopardizing recovery, all updating
transactions are aborted with internal error code 410
(Out of log file space temporarily ). This
condition prevails until a checkpoint has completed and the
log tail can be moved forward.
Important
This parameter cannot be changed “on the
fly”; you must restart the node using
--initial . If you wish to change this
value for all data nodes in a running cluster, you can do
so via a rolling node restart (using
--initial when starting each data node).
FragmentLogFileSize
Setting this parameter allows you to control directly the
size of redo log files. This can be useful in situations
when MySQL Cluster is operating under a high load and it is
unable to close fragment log files quickly enough before
attempting to open new ones (only 2 fragment log files can
be open at one time); increasing the size of the fragment
log files gives the cluster more time before having to open
each new fragment log file. The default value for this
parameter is 16M. FragmentLogFileSize was
added in MySQL Cluster NDB 6.1.11.
For more information about fragment log files, see the
description of the
NoOfFragmentLogFiles
parameter.
InitFragmentLogFiles
By default, fragment log files are created sparsely when
performing an initial start of a data node — that is,
depending on the operating system and file system in use,
not all bytes are necessarily written to disk. Beginning
with MySQL Cluster NDB 6.3.19, it is possible to override
this behavior and force all bytes to be written regardless
of the platform and file system type being used by mean of
this parameter.
InitFragmentLogFiles takes one of two
values:
Depending on your operating system and file system, setting
InitFragmentLogFiles=FULL may help
eliminate I/O errors on writes to the REDO log.
MaxNoOfOpenFiles
This parameter sets a ceiling on how many internal threads
to allocate for open files. Any situation
requiring a change in this parameter should be reported as a
bug.
The default value is 0. (Prior to MySQL 5.1.16, the default
was 40.) However, the minimum value to which this parameter
can be set is 20.
InitialNoOfOpenFiles
This parameter sets the initial number of internal threads
to allocate for open files.
The default value is 27.
MaxNoOfSavedMessages
This parameter sets the maximum number of trace files that
are kept before overwriting old ones. Trace files are
generated when, for whatever reason, the node crashes.
The default is 25 trace files.
MaxLCPStartDelay
In parallel data node recovery (supported in MySQL Cluster
NDB 6.3.8 and later), only table data is actually copied and
synchronized in parallel; synchronization of metadata such
as dictionary and checkpoint information is done in a serial
fashion. In addition, recovery of dictionary and checkpoint
information cannot be executed in parallel with performing
of local checkpoints. This means that, when starting or
restarting many data nodes concurrently, data nodes may be
forced to wait while a local checkpoint is performed, which
can result in longer node recovery times.
Beginning with MySQL Cluster NDB 6.3.23 and MySQL Cluster
NDB 6.4.3, it is possible to force a delay in the local
checkpoint to allow more (and possibly all) data nodes to
complete metadata synchronization; once each data
node's metadata synchronization is complete, all of the
data nodes can recover table data in parallel, even while
the local checkpoint is being executed.
To force such a delay, you can set
MaxLCPStartDelay , which determines the
number of seconds the cluster can wait to begin a local
checkpoint while data nodes continue to synchronize
metadata. This parameter should be set in the [ndbd
default] section of the
config.ini file, so that it is the same
for all data nodes. The maximum value is 600; the default is
0.
Metadata objects.
The next set of [ndbd] parameters defines
pool sizes for metadata objects, used to define the maximum
number of attributes, tables, indexes, and trigger objects
used by indexes, events, and replication between clusters.
Note that these act merely as “suggestions” to
the cluster, and any that are not specified revert to the
default values shown.
MaxNoOfAttributes
Defines the number of attributes that can be defined in the
cluster.
The default value is 1000, with the minimum possible value
being 32. The maximum is 4294967039. Each attribute consumes
around 200 bytes of storage per node due to the fact that
all metadata is fully replicated on the servers.
When setting MaxNoOfAttributes , it is
important to prepare in advance for any
ALTER TABLE statements that
you might want to perform in the future. This is due to the
fact, during the execution of ALTER
TABLE on a Cluster table, 3 times the number of
attributes as in the original table are used, and a good
practice is to allow double this amount. For example, if the
MySQL Cluster table having the greatest number of attributes
(greatest_number_of_attributes )
has 100 attributes, a good starting point for the value of
MaxNoOfAttributes would be 6 *
greatest_number_of_attributes =
600 .
You should also estimate the average number of attributes
per table and multiply this by
MaxNoOfTables . If this value is larger
than the value obtained in the previous paragraph, you
should use the larger value instead.
Assuming that you can create all desired tables without any
problems, you should also verify that this number is
sufficient by trying an actual ALTER
TABLE after configuring the parameter. If this is
not successful, increase
MaxNoOfAttributes by another multiple of
MaxNoOfTables and test it again.
MaxNoOfTables
A table object is allocated for each table and for each
unique hash index in the cluster. This parameter sets the
maximum number of table objects for the cluster as a whole.
For each attribute that has a
BLOB data type an extra table
is used to store most of the
BLOB data. These tables also
must be taken into account when defining the total number of
tables.
The default value of this parameter is 128. The minimum is 8
and the maximum is 20320. Each table object consumes
approximately 20KB per node.
Note
The sum of MaxNoOfTables ,
MaxNoOfOrderedIndexes , and
MaxNoOfUniqueHashIndexes must not
exceed 232 –
2 (4294967294).
MaxNoOfOrderedIndexes
For each ordered index in the cluster, an object is
allocated describing what is being indexed and its storage
segments. By default, each index so defined also defines an
ordered index. Each unique index and primary key has both an
ordered index and a hash index.
MaxNoOfOrderedIndexes sets the total
number of hash indexes that can be in use in the system at
any one time.
The default value of this parameter is 128. Each hash index
object consumes approximately 10KB of data per node.
Note
The sum of MaxNoOfTables ,
MaxNoOfOrderedIndexes , and
MaxNoOfUniqueHashIndexes must not
exceed 232 –
2 (4294967294).
MaxNoOfUniqueHashIndexes
For each unique index that is not a primary key, a special
table is allocated that maps the unique key to the primary
key of the indexed table. By default, an ordered index is
also defined for each unique index. To prevent this, you
must specify the USING HASH option when
defining the unique index.
The default value is 64. Each index consumes approximately
15KB per node.
Note
The sum of MaxNoOfTables ,
MaxNoOfOrderedIndexes , and
MaxNoOfUniqueHashIndexes must not
exceed 232 –
2 (4294967294).
MaxNoOfTriggers
Internal update, insert, and delete triggers are allocated
for each unique hash index. (This means that three triggers
are created for each unique hash index.) However, an
ordered index requires only a single
trigger object. Backups also use three trigger objects for
each normal table in the cluster.
Replication between clusters also makes use of internal
triggers.
This parameter sets the maximum number of trigger objects in
the cluster.
The default value is 768.
MaxNoOfIndexes
This parameter is deprecated in MySQL 5.1; you
should use MaxNoOfOrderedIndexes and
MaxNoOfUniqueHashIndexes instead.
This parameter is used only by unique hash indexes. There
needs to be one record in this pool for each unique hash
index defined in the cluster.
The default value of this parameter is 128.
MaxNoOfSubscriptions
Each NDB table in a MySQL
Cluster requires a subscription in the NDB kernel. For some
NDB API applications, it may be necessary or desirable to
change this parameter, which became available in MySQL
Cluster NDB 6.2.10 and MySQL Cluster NDB 6.3.7. However, for
normal usage with MySQL servers acting as SQL nodes, there
is not any need to do so.
The default value for
MaxNoOfSubscriptions is 0, which is
treated as equal to MaxNoOfTables .
MaxNoOfSubscribers
This parameter, added in MySQL Cluster NDB 6.2.10 and MySQL
Cluster NDB 6.3.7, is of interest only when using MySQL
Cluster Replication. The default value is 0, which is
treated as 2 * MaxNoOfTables ; that is,
there is one subscription per
NDB table for each of two MySQL
servers (one acting as the replication master and the other
as the slave).
When using circular replication, multi-master replcation,
and other replication setups involving more than 2 MySQL
servers, you should increase this parameter to the number of
mysqld processes included in replication
(this is often, but not always, the same as the number of
clusters). For example, if you have a circular replication
setup using three MySQL Clusters, with one
mysqld attached to each cluster, and each
of these mysqld processes acts as a
master and as a slave, you should set
MaxNoOfSubscribers equal to 3 *
MaxNoOfTables .
For more information, see
Section 17.6, “MySQL Cluster Replication”.
MaxNoOfConcurrentSubOperations
This parameter sets a ceiling on the number of operations
that can be performed by all API nodes in the cluster at one
time. The default value (256) is sufficient for normal
operations, and might need to be adjusted only in scenarios
where there are a great many API nodes each performing a
high volume of operations concurrently.
This parameter was added in MySQL Cluster NDB 6.2.10 and
MySQL Cluster NDB 6.3.7.
Boolean parameters.
The behavior of data nodes is also affected by a set of
[ndbd] parameters taking on boolean values.
These parameters can each be specified as
TRUE by setting them equal to
1 or Y , and as
FALSE by setting them equal to
0 or N .
LockPagesInMainMemory
For a number of operating systems, including Solaris and
Linux, it is possible to lock a process into memory and so
avoid any swapping to disk. This can be used to help
guarantee the cluster's real-time characteristics.
Beginning with MySQL 5.1.15 and MySQL Cluster NDB 6.1.1,
this parameter takes one of the integer values
0 , 1 , or
2 , which act as follows:
0 : Disables locking. This is the
default value.
1 : Performs the lock after allocating
memory for the process.
2 : Performs the lock before memory
for the process is allocated.
Previously, this parameter was a Boolean.
0 or false was the
default setting, and disabled locking. 1
or true enabled locking of the process
after its memory was allocated.
Important
Beginning with MySQL 5.1.15 and MySQL Cluster NDB 6.1.1,
it is no longer possible to use true or
false for the value of this parameter;
when upgrading from a previous version, you must change
the value to 0 , 1 ,
or 2 .
Prior to MySQL Cluster NDB 6.3.31 and MySQL Cluster NDB
7.0.11, setting this parameter did not cause the stated
memory to be allocated when the node was started, but rather
only when the memory was used by the data node process for
other reasons. (Bug#37430)
Note
If the operating system is not configured to allow
unprivileged users to lock pages, then the data node
process making use of this parameter may have to be run as
system root. (LockPagesInMainMemory
uses the mlockall function. From Linux
kernel 2.6.9, unprivileged users can lock memory as
limited by max locked memory . For more
information, see ulimit -l and
http://linux.die.net/man/2/mlock).
StopOnError
This parameter specifies whether an ndbd
process should exit or perform an automatic restart when an
error condition is encountered.
This feature is enabled by default.
Diskless
It is possible to specify MySQL Cluster tables as
diskless, meaning that tables are not
checkpointed to disk and that no logging occurs. Such tables
exist only in main memory. A consequence of using diskless
tables is that neither the tables nor the records in those
tables survive a crash. However, when operating in diskless
mode, it is possible to run ndbd on a
diskless computer.
Important
This feature causes the entire
cluster to operate in diskless mode.
When this feature is enabled, Cluster online backup is
disabled. In addition, a partial start of the cluster is not
possible.
Diskless is disabled by default.
ODirect
Enabling this parameter causes
NDB to attempt using
O_DIRECT writes for LCP, backups, and
redo logs, often lowering kswapd and CPU
usage. When using MySQL Cluster on Linux, enable
ODirect if you are using a 2.6 or kernel.
This parameter was added in the following releases:
MySQL 5.1.20
MySQL Cluster NDB 6.1.11
MySQL Cluster NDB 6.2.3
MySQL Cluster NDB 6.3.0
ODirect is disabled by default.
RestartOnErrorInsert
This feature is accessible only when building the debug
version where it is possible to insert errors in the
execution of individual blocks of code as part of testing.
This feature is disabled by default.
CompressedBackup
Setting this parameter to 1 causes backup
files to be compressed. The compression used is equivalent
to gzip --fast, and can save 50% or more
of the space required on the data node to store uncompressed
backup files. Compressed backups can be enabled for
individual data nodes, or for all data nodes (by setting
this parameter in the [ndbd default]
section of the config.ini file).
Important
You cannot restore a compressed backup to a cluster
running a MySQL version that does not support this
feature.
The default value is 0 (disabled).
This parameter was introduced in MySQL Cluster NDB 6.3.7.
CompressedLCP
Setting this parameter to 1 causes local
checkpoint files to be compressed. The compression used is
equivalent to gzip --fast, and can save
50% or more of the space required on the data node to store
uncompressed checkpoint files. Compressed LCPs can be
enabled for individual data nodes, or for all data nodes (by
setting this parameter in the [ndbd
default] section of the
config.ini file).
Important
You cannot restore a compressed local checkpoint to a
cluster running a MySQL version that does not support this
feature.
The default value is 0 (disabled).
This parameter was introduced in MySQL Cluster NDB 6.3.7.
Controlling Timeouts, Intervals, and Disk
Paging
There are a number of [ndbd] parameters
specifying timeouts and intervals between various actions in
Cluster data nodes. Most of the timeout values are specified in
milliseconds. Any exceptions to this are mentioned where
applicable.
TimeBetweenWatchDogCheck
To prevent the main thread from getting stuck in an endless
loop at some point, a “watchdog” thread checks
the main thread. This parameter specifies the number of
milliseconds between checks. If the process remains in the
same state after three checks, the watchdog thread
terminates it.
This parameter can easily be changed for purposes of
experimentation or to adapt to local conditions. It can be
specified on a per-node basis although there seems to be
little reason for doing so.
The default timeout is 6000 milliseconds (6 seconds).
TimeBetweenWatchDogCheckInitial
This is similar to the
TimeBetweenWatchDogCheck parameter,
except that
TimeBetweenWatchDogCheckInitial controls
the amount of time that passes between execution checks
inside a database node in the early start phases during
which memory is allocated.
The default timeout is 6000 milliseconds (6 seconds).
This parameter was added in MySQL 5.1.20.
StartPartialTimeout
This parameter specifies how long the Cluster waits for all
data nodes to come up before the cluster initialization
routine is invoked. This timeout is used to avoid a partial
Cluster startup whenever possible.
This parameter is overridden when performing an initial
start or initial restart of the cluster.
The default value is 30000 milliseconds (30 seconds). 0
disables the timeout, in which case the cluster may start
only if all nodes are available.
StartPartitionedTimeout
If the cluster is ready to start after waiting for
StartPartialTimeout milliseconds but is
still possibly in a partitioned state, the cluster waits
until this timeout has also passed. If
StartPartitionedTimeout is set to 0, the
cluster waits indefinitely.
This parameter is overridden when performing an initial
start or initial restart of the cluster.
The default timeout is 60000 milliseconds (60 seconds).
StartFailureTimeout
If a data node has not completed its startup sequence within
the time specified by this parameter, the node startup
fails. Setting this parameter to 0 (the default value) means
that no data node timeout is applied.
For nonzero values, this parameter is measured in
milliseconds. For data nodes containing extremely large
amounts of data, this parameter should be increased. For
example, in the case of a data node containing several
gigabytes of data, a period as long as 10–15 minutes
(that is, 600000 to 1000000 milliseconds) might be required
to perform a node restart.
HeartbeatIntervalDbDb
One of the primary methods of discovering failed nodes is by
the use of heartbeats. This parameter states how often
heartbeat signals are sent and how often to expect to
receive them. After missing three heartbeat intervals in a
row, the node is declared dead. Thus, the maximum time for
discovering a failure through the heartbeat mechanism is
four times the heartbeat interval.
The default heartbeat interval is 1500 milliseconds (1.5
seconds). This parameter must not be changed drastically and
should not vary widely between nodes. If one node uses 5000
milliseconds and the node watching it uses 1000
milliseconds, obviously the node will be declared dead very
quickly. This parameter can be changed during an online
software upgrade, but only in small increments.
HeartbeatIntervalDbApi
Each data node sends heartbeat signals to each MySQL server
(SQL node) to ensure that it remains in contact. If a MySQL
server fails to send a heartbeat in time it is declared
“dead,” in which case all ongoing transactions
are completed and all resources released. The SQL node
cannot reconnect until all activities initiated by the
previous MySQL instance have been completed. The
three-heartbeat criteria for this determination are the same
as described for HeartbeatIntervalDbDb .
The default interval is 1500 milliseconds (1.5 seconds).
This interval can vary between individual data nodes because
each data node watches the MySQL servers connected to it,
independently of all other data nodes.
TimeBetweenLocalCheckpoints
This parameter is an exception in that it does not specify a
time to wait before starting a new local checkpoint; rather,
it is used to ensure that local checkpoints are not
performed in a cluster where relatively few updates are
taking place. In most clusters with high update rates, it is
likely that a new local checkpoint is started immediately
after the previous one has been completed.
The size of all write operations executed since the start of
the previous local checkpoints is added. This parameter is
also exceptional in that it is specified as the base-2
logarithm of the number of 4-byte words, so that the default
value 20 means 4MB (4 ?
220) of write operations, 21
would mean 8MB, and so on up to a maximum value of 31, which
equates to 8GB of write operations.
All the write operations in the cluster are added together.
Setting TimeBetweenLocalCheckpoints to 6
or less means that local checkpoints will be executed
continuously without pause, independent of the cluster's
workload.
TimeBetweenGlobalCheckpoints
When a transaction is committed, it is committed in main
memory in all nodes on which the data is mirrored. However,
transaction log records are not flushed to disk as part of
the commit. The reasoning behind this behavior is that
having the transaction safely committed on at least two
autonomous host machines should meet reasonable standards
for durability.
It is also important to ensure that even the worst of cases
— a complete crash of the cluster — is handled
properly. To guarantee that this happens, all transactions
taking place within a given interval are put into a global
checkpoint, which can be thought of as a set of committed
transactions that has been flushed to disk. In other words,
as part of the commit process, a transaction is placed in a
global checkpoint group. Later, this group's log records are
flushed to disk, and then the entire group of transactions
is safely committed to disk on all computers in the cluster.
This parameter defines the interval between global
checkpoints. The default is 2000 milliseconds.
TimeBetweenEpochs
This parameter defines the interval between synchronisation
epochs for MySQL Cluster Replication. The default value is
100 milliseconds.
TimeBetweenEpochs is part of the
implementation of “micro-GCPs”, which can be
used to improve the performance of MySQL Cluster
Replication. This parameter was introduced in MySQL Cluster
NDB 6.2.5 and MySQL Cluster NDB 6.3.2.
TimeBetweenEpochsTimeout
This parameter defines a timeout for synchronisation epochs
for MySQL Cluster Replication. If a node fails to
participate in a global checkpoint within the time
determined by this parameter, the node is shut down. The
default value is 4000 milliseconds.
TimeBetweenEpochsTimeout is part of the
implementation of “micro-GCPs”, which can be
used to improve the performance of MySQL Cluster
Replication. This parameter was introduced in MySQL Cluster
NDB 6.2.7 and MySQL Cluster NDB 6.3.4.
MaxBufferedEpochs
The number of unprocessed epochs by which a subscribing node
can lag behind. Exceeding this number causes a lagging
subscriber to be disconnected.
The default value of 100 is sufficient for most normal
operations. If a subscribing node does lag enough to cause
disconnections, it is usually due to network or scheduling
issues with regard to processes or threads. (In rare
circumstances, the problem may be due to a bug in the
NDB client.) It may be
desirable to set the value lower than the default when
epochs are longer.
Disconnection prevents client issues from affecting the data
node service, running out of memory to buffer data, and
eventually shutting down. Instead, only the client is
affected as a result of the disconnect (by, for example gap
events in the binlog), forcing the client to reconnect or
restart the process.
TimeBetweenInactiveTransactionAbortCheck
Timeout handling is performed by checking a timer on each
transaction once for every interval specified by this
parameter. Thus, if this parameter is set to 1000
milliseconds, every transaction will be checked for timing
out once per second.
The default value is 1000 milliseconds (1 second).
TransactionInactiveTimeout
This parameter states the maximum time that is permitted to
lapse between operations in the same transaction before the
transaction is aborted.
The default for this parameter is zero (no timeout). For a
real-time database that needs to ensure that no transaction
keeps locks for too long, this parameter should be set to a
relatively small value. The unit is milliseconds.
TransactionDeadlockDetectionTimeout
When a node executes a query involving a transaction, the
node waits for the other nodes in the cluster to respond
before continuing. A failure to respond can occur for any of
the following reasons:
This timeout parameter states how long the transaction
coordinator waits for query execution by another node before
aborting the transaction, and is important for both node
failure handling and deadlock detection. In MySQL 5.1.10 and
earlier versions, setting it too high could cause
undesirable behavior in situations involving deadlocks and
node failure. Beginning with MySQL 5.1.11, active
transactions occurring during node failures are actively
aborted by the MySQL Cluster Transaction Coordinator, and so
high settings are no longer an issue with this parameter.
The default timeout value is 1200 milliseconds (1.2
seconds).
Prior to MySQL Cluster NDB versions 6.2.18, 6.3.24, and
7.0.5, the effective minimum for this parameter was 100
milliseconds. (Bug#44099) Beginning with these versions, the
actual minimum is 50 milliseconds.
DiskSyncSize
This is the maximum number of bytes to store before flushing
data to a local checkpoint file. This is done in order to
prevent write buffering, which can impede performance
significantly. This parameter is not
intended to take the place of
TimeBetweenLocalCheckpoints .
Note
When ODirect is enabled, it is not
necessary to set DiskSyncSize ; in fact,
in such cases its value is simply ignored.
The default value is 4M (4 megabytes).
This parameter was added in MySQL 5.1.12.
DiskCheckpointSpeed
The amount of data,in bytes per second, that is sent to disk
during a local checkpoint. This allocation is shared by DML
operations and backups (but not backup logging), which means
that backups started during times of intensive DML may be
impaired by flooding of the redo log buffer and may fail
altogether if the contention is sufficiently severe.
The default value is 10M (10 megabytes per second).
This parameter was added in MySQL 5.1.12.
DiskCheckpointSpeedInRestart
The amount of data,in bytes per second, that is sent to disk
during a local checkpoint as part of a restart operation.
The default value is 100M (100 megabytes per second).
This parameter was added in MySQL 5.1.12.
NoOfDiskPagesToDiskAfterRestartTUP
When executing a local checkpoint, the algorithm flushes all
data pages to disk. Merely doing so as quickly as possible
without any moderation is likely to impose excessive loads
on processors, networks, and disks. To control the write
speed, this parameter specifies how many pages per 100
milliseconds are to be written. In this context, a
“page” is defined as 8KB. This parameter is
specified in units of 80KB per second, so setting
NoOfDiskPagesToDiskAfterRestartTUP to a
value of 20 entails writing 1.6MB in data
pages to disk each second during a local checkpoint. This
value includes the writing of UNDO log records for data
pages. That is, this parameter handles the limitation of
writes from data memory. (See the entry for
IndexMemory for information about index
pages.)
In short, this parameter specifies how quickly to execute
local checkpoints. It operates in conjunction with
NoOfFragmentLogFiles ,
DataMemory , and
IndexMemory .
For more information about the interaction between these
parameters and possible strategies for choosing appropriate
values for them, see
Section 17.3.2.12, “Configuring MySQL Cluster Parameters for Local Checkpoints”.
The default value is 40 (3.2MB of data pages per second).
NoOfDiskPagesToDiskAfterRestartACC
This parameter uses the same units as
NoOfDiskPagesToDiskAfterRestartTUP and
acts in a similar fashion, but limits the speed of writing
index pages from index memory.
The default value of this parameter is 20 (1.6MB of index
memory pages per second).
NoOfDiskPagesToDiskDuringRestartTUP
This parameter is used in a fashion similar to
NoOfDiskPagesToDiskAfterRestartTUP and
NoOfDiskPagesToDiskAfterRestartACC , only
it does so with regard to local checkpoints executed in the
node when a node is restarting. A local checkpoint is always
performed as part of all node restarts. During a node
restart it is possible to write to disk at a higher speed
than at other times, because fewer activities are being
performed in the node.
This parameter covers pages written from data memory.
The default value is 40 (3.2MB per second).
NoOfDiskPagesToDiskDuringRestartACC
Controls the number of index memory pages that can be
written to disk during the local checkpoint phase of a node
restart.
As with
NoOfDiskPagesToDiskAfterRestartTUP and
NoOfDiskPagesToDiskAfterRestartACC ,
values for this parameter are expressed in terms of 8KB
pages written per 100 milliseconds (80KB/second).
The default value is 20 (1.6MB per second).
ArbitrationTimeout
This parameter specifies how long data nodes wait for a
response from the arbitrator to an arbitration message. If
this is exceeded, the network is assumed to have split.
The default value is 1000 milliseconds (1 second).
Arbitration
The Arbitration parameter, added in MySQL
Cluster NDB 7.0.7, allows a choice of arbitration schemes,
corresponding to one of 3 possible values for this
parameter:
Default .
This allows arbitration to proceed normally, as
determined by the ArbitrationRank
settings for the management and API nodes. This is the
default value.
Disabled .
Previously, it was possible to disable arbitration
only by setting ArbitrationRank to
0 on all management and API nodes. Now, you can now
use Arbitration = Disabled in the
[ndbd default] section of the
config.ini file to accomplish
this task. In this case, any
ArbitrationRank settings are
ignored.
WaitExternal .
The Arbitration parameter also
makes it possible to configure arbitration in such a
way that the cluster waits until after the time
determined by ArbitrationTimeout
has passed for an external cluster manager application
to perform arbitration instead of handling arbitration
internally. This can be done by setting
Arbitration = WaitExternal in the
[ndbd default] section of the
config.ini file. For best results
with the WaitExternal setting, it
is recommended that
ArbitrationTimeout be 2 times as
long as the interval required by the external cluster
manager to perform arbitration.
Important
This parameter should be used only in the [ndbd
default] section of the cluster configuration
file. The behavior of the cluster is unspecified when
Arbitration is set to different values
for individual data nodes.
Buffering and logging.
Several [ndbd] configuration parameters
enable the advanced user to have more control over the
resources used by node processes and to adjust various buffer
sizes at need.
These buffers are used as front ends to the file system when
writing log records to disk. If the node is running in diskless
mode, these parameters can be set to their minimum values
without penalty due to the fact that disk writes are
“faked” by the NDB
storage engine's file system abstraction layer.
UndoIndexBuffer
The UNDO index buffer, whose size is set by this parameter,
is used during local checkpoints. The
NDB storage engine uses a
recovery scheme based on checkpoint consistency in
conjunction with an operational REDO log. To produce a
consistent checkpoint without blocking the entire system for
writes, UNDO logging is done while performing the local
checkpoint. UNDO logging is activated on a single table
fragment at a time. This optimization is possible because
tables are stored entirely in main memory.
The UNDO index buffer is used for the updates on the primary
key hash index. Inserts and deletes rearrange the hash
index; the NDB storage engine writes UNDO log records that
map all physical changes to an index page so that they can
be undone at system restart. It also logs all active insert
operations for each fragment at the start of a local
checkpoint.
Reads and updates set lock bits and update a header in the
hash index entry. These changes are handled by the
page-writing algorithm to ensure that these operations need
no UNDO logging.
This buffer is 2MB by default. The minimum value is 1MB,
which is sufficient for most applications. For applications
doing extremely large or numerous inserts and deletes
together with large transactions and large primary keys, it
may be necessary to increase the size of this buffer. If
this buffer is too small, the NDB storage engine issues
internal error code 677 (Index UNDO buffers
overloaded ).
Important
It is not safe to decrease the value of this parameter
during a rolling restart.
UndoDataBuffer
This parameter sets the size of the UNDO data buffer, which
performs a function similar to that of the UNDO index
buffer, except the UNDO data buffer is used with regard to
data memory rather than index memory. This buffer is used
during the local checkpoint phase of a fragment for inserts,
deletes, and updates.
Because UNDO log entries tend to grow larger as more
operations are logged, this buffer is also larger than its
index memory counterpart, with a default value of 16MB.
This amount of memory may be unnecessarily large for some
applications. In such cases, it is possible to decrease this
size to a minimum of 1MB.
It is rarely necessary to increase the size of this buffer.
If there is such a need, it is a good idea to check whether
the disks can actually handle the load caused by database
update activity. A lack of sufficient disk space cannot be
overcome by increasing the size of this buffer.
If this buffer is too small and gets congested, the NDB
storage engine issues internal error code 891
(Data UNDO buffers overloaded).
Important
It is not safe to decrease the value of this parameter
during a rolling restart.
RedoBuffer
All update activities also need to be logged. The REDO log
makes it possible to replay these updates whenever the
system is restarted. The NDB recovery algorithm uses a
“fuzzy” checkpoint of the data together with
the UNDO log, and then applies the REDO log to play back all
changes up to the restoration point.
RedoBuffer sets the size of the buffer in
which the REDO log is written. In MySQL Cluster NDB 6.4.3
and earlier, the default value is 8MB; beginning with MySQL
Cluster NDB 7.0.4, the default is 32MB. The minimum value is
1MB.
If this buffer is too small, the NDB storage engine issues
error code 1221 (REDO log buffers
overloaded ).
Important
It is not safe to decrease the value of this parameter
during a rolling restart.
Controlling log messages.
In managing the cluster, it is very important to be able to
control the number of log messages sent for various event
types to stdout . For each event category,
there are 16 possible event levels (numbered 0 through 15).
Setting event reporting for a given event category to level 15
means all event reports in that category are sent to
stdout ; setting it to 0 means that there
will be no event reports made in that category.
By default, only the startup message is sent to
stdout , with the remaining event reporting
level defaults being set to 0. The reason for this is that these
messages are also sent to the management server's cluster log.
An analogous set of levels can be set for the management client
to determine which event levels to record in the cluster log.
LogLevelStartup
The reporting level for events generated during startup of
the process.
The default level is 1.
LogLevelShutdown
The reporting level for events generated as part of graceful
shutdown of a node.
The default level is 0.
LogLevelStatistic
The reporting level for statistical events such as number of
primary key reads, number of updates, number of inserts,
information relating to buffer usage, and so on.
The default level is 0.
LogLevelCheckpoint
The reporting level for events generated by local and global
checkpoints.
The default level is 0.
LogLevelNodeRestart
The reporting level for events generated during node
restart.
The default level is 0.
LogLevelConnection
The reporting level for events generated by connections
between cluster nodes.
The default level is 0.
LogLevelError
The reporting level for events generated by errors and
warnings by the cluster as a whole. These errors do not
cause any node failure but are still considered worth
reporting.
The default level is 0.
LogLevelCongestion
The reporting level for events generated by congestion.
These errors do not cause node failure but are still
considered worth reporting.
The default level is 0.
LogLevelInfo
The reporting level for events generated for information
about the general state of the cluster.
The default level is 0.
MemReportFrequency
This parameter controls how often data node memory usage
reports are recorded in the cluster log; it is an integer
value representing the number of seconds between reports.
Each data node's data memory and index memory usage is
logged as both a percentage and a number of 32 KB pages of
the DataMemory and
IndexMemory , respectively, set in the
config.ini file. For example, if
DataMemory is equal to 100 MB, and a
given data node is using 50 MB for data memory storage, the
corresponding line in the cluster log might look like this:
2006-12-24 01:18:16 [MgmSrvr] INFO -- Node 2: Data usage is 50%(1280 32K pages of total 2560)
MemReportFrequency is not a required
parameter. If used, it can be set for all cluster data nodes
in the [ndbd default] section of
config.ini , and can also be set or
overridden for individual data nodes in the corresponding
[ndbd] sections of the configuration
file. The minimum value — which is also the default
value — is 0, in which case memory reports are logged
only when memory usage reaches certain percentages (80%,
90%, and 100%), as mentioned in the discussion of statistics
events in Section 17.5.4.2, “MySQL Cluster Log Events”.
This parameter was added in MySQL Cluster 5.1.16 and MySQL
Cluster NDB 6.1.0.
StartupStatusReportFrequency
When a data node is started with the
--initial , it initializes the
redo log file during Start Phase 4 (see
Section 17.5.1, “Summary of MySQL Cluster Start Phases”). When very
large values are set for
NoOfFragmentLogFiles ,
FragmentLogFileSize , or both, this
initialization can take a long time. Previous to MySQL
Cluster NDB 6.4.0, only the beginning and end of the redo
log file initialization process were logged. Beginning with
this version, it is possible to force reports on the
progress of this process to be logged periodically, by means
of the StartupStatusReportFrequency
configuration parameter. In this case, progress is reported
in the cluster log, in terms of both the number of files and
the amount of space that have been initialized, as shown
here:
2009-06-20 16:39:23 [MgmSrvr] INFO -- Node 1: Local redo log file initialization status:
#Total files: 80, Completed: 60
#Total MBytes: 20480, Completed: 15557
2009-06-20 16:39:23 [MgmSrvr] INFO -- Node 2: Local redo log file initialization status:
#Total files: 80, Completed: 60
#Total MBytes: 20480, Completed: 15570
These reports are logged each
StartupStatusReportFrequency seconds
during Start Pahe 4. If
StartupStatusReportFrequency is 0 (the
default), then reports are written to the cluster log only
when at the beginning and at the completion of the redo log
file initialization process.
Backup parameters.
The [ndbd] parameters discussed in this
section define memory buffers set aside for execution of
online backups.
BackupDataBufferSize
In creating a backup, there are two buffers used for sending
data to the disk. The backup data buffer is used to fill in
data recorded by scanning a node's tables. Once this buffer
has been filled to the level specified as
BackupWriteSize (see below), the pages
are sent to disk. While flushing data to disk, the backup
process can continue filling this buffer until it runs out
of space. When this happens, the backup process pauses the
scan and waits until some disk writes have completed freed
up memory so that scanning may continue.
In MySQL Cluster NDB 6.4.3 and earlier, the default value is
2MB; in MySQL Cluster NDB 7.0.4 and later, it is 16MB.
BackupLogBufferSize
The backup log buffer fulfills a role similar to that played
by the backup data buffer, except that it is used for
generating a log of all table writes made during execution
of the backup. The same principles apply for writing these
pages as with the backup data buffer, except that when there
is no more space in the backup log buffer, the backup fails.
For that reason, the size of the backup log buffer must be
large enough to handle the load caused by write activities
while the backup is being made. See
Section 17.5.3.3, “Configuration for MySQL Cluster Backups”.
The default value for this parameter should be sufficient
for most applications. In fact, it is more likely for a
backup failure to be caused by insufficient disk write speed
than it is for the backup log buffer to become full. If the
disk subsystem is not configured for the write load caused
by applications, the cluster is unlikely to be able to
perform the desired operations.
It is preferable to configure cluster nodes in such a manner
that the processor becomes the bottleneck rather than the
disks or the network connections.
In MySQL Cluster NDB 6.4.3 and earlier, the default value is
2MB; in MySQL Cluster NDB 7.0.4 and later, it is 16MB.
BackupMemory
This parameter is simply the sum of
BackupDataBufferSize and
BackupLogBufferSize .
In MySQL Cluster NDB 6.4.3 and earlier, the default value
was 2MB + 2MB = 4MB; in MySQL Cluster NDB 7.0.4 and later,
it is 16MB + 16MB = 32MB.
Important
If BackupDataBufferSize and
BackupLogBufferSize taken together
exceed the default value for
BackupMemory , then this parameter must
be set explicitly in the config.ini
file to their sum.
BackupReportFrequency
This parameter controls how often backup status reports are
issued in the management client during a backup, as well as
how often such reports are written to the cluster log
(provided cluster event logging is configured to allow it
— see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”).
BackupReportFrequency represents the time
in seconds between backup status reports.
The default value is 0.
This parameter was added in MySQL Cluster NDB 6.2.3.
BackupWriteSize
This parameter specifies the default size of messages
written to disk by the backup log and backup data buffers.
In MySQL Cluster 6.4.3 and earlier, the default value for
this parameter was 32KB; beginning with MySQL Cluster NDB
7.0.4, it is 256KB.
BackupMaxWriteSize
This parameter specifies the maximum size of messages
written to disk by the backup log and backup data buffers.
In MySQL Cluster 6.4.3 and earlier, the default value for
this parameter was 256KB; beginning with MySQL Cluster NDB
7.0.4, it is 1MB.
Important
When specifying these parameters, the following relationships
must hold true. Otherwise, the data node will be unable to
start.
BackupDataBufferSize >= BackupWriteSize +
188KB
BackupLogBufferSize >= BackupWriteSize +
16KB
BackupMaxWriteSize >= BackupWriteSize
Realtime Performance Parameters
The [ndbd] parameters discussed in this
section are used in scheduling and locking of threads to
specific CPUs on multiprocessor data node hosts. They were
introduced in MySQL Cluster NDB 6.3.4.
Note
To make use of these parameters, the data node process must be
run as system root.
LockExecuteThreadToCPU
Previous to MySQL Cluster NDB 7.0.
This parameter specifies the ID of the CPU assigned to
handle the NDBCLUSTER
execution thread. The value of this parameter is an
integer in the range 0 to 65535 (inclusive). The default
is 65535.
MySQL Cluster NDB 7.0 and later (beginning with MySQL Cluster NDB 6.4.0).
When used with ndbd, this parameter
(now a string) specifies the ID of the CPU assigned to
handle the NDBCLUSTER
execution thread. When used with
ndbmtd, the value of this parameter is
a comma-separated list of CPU IDs assigned to handle
execution threads. Each CPU ID in the list should be an
integer in the range 0 to 65535 (inclusive). The number of
IDs specified should match the number of execution threads
determined by MaxNoOfExecutionThreads .
There is no default value.
LockMaintThreadsToCPU
This parameter specifies the ID of the CPU assigned to
handle NDBCLUSTER maintenance
threads.
The value of this parameter is an integer in the range 0 to
65535 (inclusive). This parameter was added in MySQL Cluster
NDB 6.3.4. Prior to MySQL Cluster NDB 6.4.0, the default is
65535; in MySQL Cluster NDB 7.0 and later MySQL Cluster
release series, there is no default value.
RealtimeScheduler
Setting this parameter to 1 enables real-time scheduling of
NDBCLUSTER threads.
The default is 0 (scheduling disabled).
SchedulerExecutionTimer
This parameter specifies the time in microseconds for
threads to be executed in the scheduler before being sent.
Setting it to 0 minimizes the response time; to achieve
higher throughput, you can increase the value at the expense
of longer response times.
The default is 50 ?sec, which our testing shows to
increase throughput slightly in high-load cases without
materially delaying requests.
This parameter was added in MySQL Cluster NDB 6.3.4.
SchedulerSpinTimer
This parameter specifies the time in microseconds for
threads to be executed in the scheduler before sleeping.
The default value is 0.
Disk Data Configuration Parameters.
Configuration parameters affecting Disk Data behavior include
the following:
DiskPageBufferMemory
This determines the amount of space used for caching
pages on disk, and is set in the
[ndbd] or [ndbd
default] section of the
config.ini file. It is measured in
bytes. Each page takes up 32 KB. This means that Cluster
Disk Data storage always uses
N * 32 KB memory where
N is some nonnegative
integer.
The default value for this parameter is
64M (2000 pages of 32 KB each).
This parameter was added in MySQL 5.1.6.
SharedGlobalMemory
This determines the amount of memory that is used for
log buffers, disk operations (such as page requests and
wait queues), and metadata for tablespaces, log file
groups, UNDO files, and data files.
It can be set in the [ndbd] or
[ndbd default] section of the
config.ini configuration file, and
is measured in bytes.
The default value is 20M .
This parameter was added in MySQL 5.1.6.
DiskIOThreadPool
This parameter determines the number of unbound threads
used for Disk Data file access. Before
DiskIOThreadPool was introduced,
exactly one thread was spawned for each Disk Data file,
which could lead to performance issues, particularly
when using very large data files. With
DiskIOThreadPool , you can — for
example — access a single large data file using
several threads working in parallel.
Currently, this parameter applies to Disk Data I/O
threads only, but we plan in the future to make the
number of such threads configurable for in-memory data
as well.
The optimum value for this parameter depends on your
hardware and configuration, and includes these factors:
Physical distribution of Disk Data files.
You can obtain better performance by placing data
files, undo log files, and the data node
filesystem on separate physical disks. If you do
this with some or all of these sets of files, then
you can set DiskIOThreadPool
higher to allow separate threads to handle the
files on each disk.
Disk performance and types.
The number of threads that can be accommodated for
Disk Data file handling is also dependent on the
speed and throughput of the disks. Faster disks
and higher throughput allow for more disk I/O
threads. Our test results indicate that
solid-state disk drives can handle many more disk
I/O threads than conventional disks, and thus
higher values for
DiskIOThreadPool .
This parameter was added in MySQL Cluster NDB 6.4.0.
Previous to MySQL Cluster NDB 6.4.3, it was named
ThreadPool . Previous to MySQL Cluster
NDB 7.0.7, the default value was 8. Beginning with MySQL
Cluster NDB 7.0.7 and MySQL Cluster NDB 7.1.0, the
default is 2.
Disk Data filesystem parameters.
The parameters in the following list were added in
MySQL Cluster NDB 6.2.17, 6.3.22, and 6.4.3 to make it
possible to place MySQL Cluster Disk Data files in
specific directories without the need for using
symbolic links.
FileSystemPathDD
If this parameter is specified, then MySQL Cluster
Disk Data data files and undo log files are placed
in the indicated directory. This can be overridden
for data files, undo log files, or both, by
specifying values for
FileSystemPathDataFiles ,
FileSystemPathUndoFiles , or both,
as explained for these parameters. It can also be
overridden for data files by specifying a path in
the ADD DATAFILE clause of a
CREATE TABLESPACE or
ALTER TABLESPACE
statement, and for undo log files by specifying a
path in the ADD UNDOFILE clause
of a CREATE LOGFILE
GROUP or ALTER
LOGFILE GROUP statement. If
FileSystemPathDD is not
specified, then FileSystemPath is
used.
If a FileSystemPathDD directory
is specified for a given data node (including the
case where the parameter is specified in the
[ndbd default] section of the
config.ini file), then starting
that data node with --initial
causes all files in the directory to be deleted.
FileSystemPathDataFiles
If this parameter is specified, then MySQL Cluster
Disk Data data files are placed in the indicated
directory. This overrides any value set for
FileSystemPathDD . This parameter
can be overridden for a given data file by
specifying a path in the ADD
DATAFILE clause of a
CREATE TABLESPACE or
ALTER TABLESPACE
statement used to create that data file. If
FileSystemPathDataFiles is not
specified, then FileSystemPathDD
is used (or FileSystemPath , if
FileSystemPathDD has also not
been set).
If a FileSystemPathDataFiles
directory is specified for a given data node
(including the case where the parameter is specified
in the [ndbd default] section of
the config.ini file), then
starting that data node with
--initial causes all files in the
directory to be deleted.
FileSystemPathUndoFiles
If this parameter is specified, then MySQL Cluster
Disk Data undo log files are placed in the indicated
directory. This overrides any value set for
FileSystemPathDD . This parameter
can be overridden for a given data file by
specifying a path in the ADD UNDO
clause of a CREATE LOGFILE
GROUP or CREATE
LOGFILE GROUP statement used to create
that data file. If
FileSystemPathUndoFiles is not
specified, then FileSystemPathDD
is used (or FileSystemPath , if
FileSystemPathDD has also not
been set).
If a FileSystemPathUndoFiles
directory is specified for a given data node
(including the case where the parameter is specified
in the [ndbd default] section of
the config.ini file), then
starting that data node with
--initial causes all files in the
directory to be deleted.
For more information, see
Section 17.5.10.1, “MySQL Cluster Disk Data Objects”.
Disk Data object creation parameters.
The next two parameters enable you — when
starting the cluster for the first time — to
cause a Disk Data log file group, tablespace, or both,
to be created without the use of SQL statements.
InitialLogFileGroup
This parameter can be used to specify a log file
group that is created when performing an initial
start of the cluster.
InitialLogFileGroup is specified
as shown here:
InitialLogFileGroup = [name=name ;] [undo_buffer_size=size ;] file-specification-list
file-specification-list :
file-specification [; file-specification [; ...]]
file-specification :
filename :size
The name of the log file group is
optional and defaults to
DEFAULT_LG . The
undo_buffer_size is also
optional; if omitted, it defaults to
256M (256 megabytes). Each
file-specification
corresponds to an undo log file, and at least one
must be specified in the
file-specification-list .
Undo log files are placed according to any values
that have been set for
FileSystemPath ,
FileSystemPathDD , and
FileSystemPathUndoFiles , just as
if they had been created as the result of a
CREATE LOGFILE GROUP
or ALTER LOGFILE
GROUP statement.
Consider the following example:
InitialLogFileGroup = name=LG1; undo_buffer_size=128M; undo1.log:250M; undo2.log:150M
This is equivalent to the following SQL statements:
CREATE LOGFILE GROUP LG1
ADD UNDOFILE 'undo1.log'
INITIAL_SIZE 250M
UNDO_BUFFER_SIZE 128M
ENGINE NDBCLUSTER;
ALTER LOGFILE GROUP LG1
ADD UNDOFILE 'undo2.log'
INITIAL_SIZE 150M
ENGINE NDBCLUSTER;
This logfile group is created when the data nodes
are started with --initial .
This parameter, if used, should always be set in the
[ndbd default] section of the
config.ini file. The behavior
of a MySQL Cluster when different values are set on
different data nodes is not defined.
InitialTablespace
This parameter can be used to specify a MySQL
Cluster Disk Data tablespace that is created when
performing an initial start of the cluster.
InitialTablespace is specified as
shown here:
InitialTablespace = [name=name ;] [extent_size=size ;] file-specification-list
The name of the tablespace is
optional and defaults to
DEFAULT_TS . The
extent_size is also optional; it
defaults to 1M (1 megabyte). The
file-specification-list
uses the same syntax as shown with the
InitialLogfileGroup parameter,
the only difference being that each
file-specification used
with InitialTablespace
corresponds to a data file. At least one must be
specified in the
file-specification-list .
Data files are placed according to any values that
have been set for FileSystemPath ,
FileSystemPathDD , and
FileSystemPathDataFiles , just as
if they had been created as the result of a
CREATE TABLESPACE or
ALTER TABLESPACE
statement.
For example, consider the following line specifying
InitialTablespace in the
[ndbd default] section of the
config.ini file (as with
InitialLogfileGroup , this
parameter should always be set in the [ndbd
default] section, as the behavior of a
MySQL Cluster when different values are set on
different data nodes is not defined):
InitialTablespace = name=TS1; extent_size=8M; data1.dat:2G; data2.dat:4G
This is equivalent to the following SQL statements:
CREATE TABLESPACE TS1
ADD DATAFILE 'data1.dat'
EXTENT_SIZE 8M
INITIAL_SIZE 2G
ENGINE NDBCLUSTER;
ALTER TABLESPACE TS1
ADD UNDOFILE 'data2.dat'
INITIAL_SIZE 4G
ENGINE NDBCLUSTER;
This tablespace is created when the data nodes are
started with --initial , and can be
used whenever creating MySQL Cluster Disk Data
tables thereafter.
Disk Data and GCP Stop errors.
Errors encountered when using Disk Data tables such as
Node nodeid killed this
node because GCP stop was detected (error 2303)
are often referred to as “GCP stop errors”. Such
errors occur when the redo log is not flushed to disk quickly
enough; this is usually due to slow disks and insufficient
disk throughput.
You can help prevent these errors from occurring by using faster
disks, and by placing Disk Data files on a separate disk from
the data node filesystem. Reducing the value of
TimeBetweenGlobalCheckpoints tends to
decrease the amount of data to be written for each global
checkpoint, and so may provide some protection against redo log
buffer overflows when trying to write a global checkpoint;
however, reducing this value also allows less time in which to
write the GCP, so this must be done with caution.
In addition, adjusting the cluster configuration as discussed
here can also help:
MySQL Cluster NDB 6.2 and 6.3.
When working with large amounts of data on disk under high
load, the default value for
DiskPageBufferMemory may not be large
enough. In such cases, you should increase its value to
include most of the memory available to the data nodes
after accounting for index memory, data memory, internal
buffers, and memory needed by the data node host operating
system.
You can use this formula as a guide:
DiskPageBufferMemory
= 0.8
x (
[total memory]
- ([operating system memory] + [buffer memory] + DataMemory + IndexMemory)
)
Once you have established that sufficient memory is reserved
for DataMemory ,
IndexMemory , NDB internal buffers, and
operating system overhead, it is possible (and sometimes
desirable) to allocate more than the above amount of the
remainder to DiskPageBufferMemory .
MySQL Cluster NDB 7.X.
In addition to the considerations given for
DiskPageBufferMemory as explained in
the previous item, it is also very important that the
DiskIOThreadPool configuration
parameter be set correctly; having
DiskIOThreadPool set too high is very
likely to cause GCP stop errors (Bug#37227).
Parameters for configuring send buffer memory allocation (MySQL Cluster
NDB 7.0).
Beginning with MySQL Cluster NDB 6.4.0, send buffer memory is
allocated dynamically from a memory pool shared between all
transporters, which means that the size of the send buffer can
be adjusted as necessary. (Previously, the NDB kernel used a
fixed-size send buffer for every node in the cluster, which
was allocated when the node started and could not be changed
while the node was running.) The following data node
configuration parameters were added in MySQL Cluster NDB 6.4.0
to permit the setting of limits on this memory allocation;
this change is reflected by the addition of the configuration
parameters TotalSendBufferMemory and
OverLoadLimit , as well as a change in how
the existing SendBufferMemory configuration
parameter is used. For more information, see
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”.
TotalSendBufferMemory
This parameter is available beginning with MySQL Cluster NDB
6.4.0. It is used to determine the total amount of memory to
allocate on this node for shared send buffer memory among
all configured transporters.
If this parameter is set, its minimum allowed value is 256K;
the maxmimum is 4294967039.
ReservedSendBufferMemory
This parameter is present in
NDBCLUSTER source code
beginning with MySQL Cluster NDB 6.4.0. However, it is not
currently enabled.
For more detailed information about the behavior and use of
TotalSendBufferMemory
and about configuring send buffer memory parameters in MySQL
Cluster NDB 6.4.0 and later, see
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”.
Note
Previous to MySQL Cluster NDB 7.0, to add new data nodes to a
MySQL Cluster, it is necessary to shut down the cluster
completely, update the config.ini file,
and then restart the cluster (that is, you must perform a
system restart). All data node processes must be started with
the --initial option.
Beginning with MySQL Cluster NDB 7.0, it is possible to add
new data node groups to a running cluster online. See
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”, for more
information.
17.3.2.7. Defining SQL and Other API Nodes in a MySQL Cluster
The [mysqld] and [api]
sections in the config.ini file define the
behavior of the MySQL servers (SQL nodes) and other applications
(API nodes) used to access cluster data. None of the parameters
shown is required. If no computer or host name is provided, any
host can use this SQL or API node.
Generally speaking, a [mysqld] section is
used to indicate a MySQL server providing an SQL interface to
the cluster, and an [api] section is used for
applications other than mysqld processes
accessing cluster data, but the two designations are actually
synonomous; you can, for instance, list parameters for a MySQL
server acting as an SQL node in an [api]
section.
Id
The Id is an integer value used to
identify the node in all cluster internal messages. Prior to
MySQL Cluster NDB 6.1.1, the permitted range of values for
this parameter was 1 to 63 inclusive. Beginning with MySQL
Cluster NDB 6.1.1, the permitted range is 1 to 255
inclusive. This value must be unique for each node in the
cluster, regardless of the type of node.
Note
Data node IDs must be less than 49, regardless of the
MySQL Cluster version used. If you plan to deploy a large
number of data nodes, it is a good idea to limit the node
IDs for API nodes (and management nodes) to values greater
than 48.
This parameter can also be written as
NodeId , although the short form is
sufficient (and preferred for this reason).
ExecuteOnComputer
This refers to the Id set for one of the
computers (hosts) defined in a [computer]
section of the configuration file.
HostName
Specifying this parameter defines the hostname of the
computer on which the SQL node (API node) is to reside. To
specify a hostname, either this parameter or
ExecuteOnComputer is required.
If no HostName or
ExecuteOnComputer is specified in a given
[mysql] or [api]
section of the config.ini file, then an
SQL or API node may connect using the corresponding
“slot” from any host which can establish a
network connection to the management server host machine.
This differs from the default behavior for data
nodes, where localhost is assumed for
HostName unless otherwise
specified.
ArbitrationRank
This parameter defines which nodes can act as arbitrators.
Both MGM nodes and SQL nodes can be arbitrators. A value of
0 means that the given node is never used as an arbitrator,
a value of 1 gives the node high priority as an arbitrator,
and a value of 2 gives it low priority. A normal
configuration uses the management server as arbitrator,
setting its ArbitrationRank to 1 (the
default for management nodes) and those for all SQL nodes to
0 (the default for SQL nodes).
Beginning with MySQL 5.1.16 and MySQL Cluster NDB 6.1.3, it
is possible to disable arbitration completely by setting
ArbitrationRank to 0 on all management
and SQL nodes. In MySQL Cluster NDB 7.0.7 and later
releases, you can also control arbitration by overriding
this parameter; to do this, set the
Arbitration
parameter in the [ndbd default] section
of the config.ini global configuration
file.
ArbitrationDelay
Setting this parameter to any other value than 0 (the
default) means that responses by the arbitrator to
arbitration requests will be delayed by the stated number of
milliseconds. It is usually not necessary to change this
value.
BatchByteSize
For queries that are translated into full table scans or
range scans on indexes, it is important for best performance
to fetch records in properly sized batches. It is possible
to set the proper size both in terms of number of records
(BatchSize ) and in terms of bytes
(BatchByteSize ). The actual batch size is
limited by both parameters.
The speed at which queries are performed can vary by more
than 40% depending upon how this parameter is set. In future
releases, MySQL Server will make educated guesses on how to
set parameters relating to batch size, based on the query
type.
This parameter is measured in bytes and by default is equal
to 32KB.
BatchSize
This parameter is measured in number of records and is by
default set to 64. The maximum size is 992.
HeartbeatThreadPriority
Beginning with MySQL Cluster NDB 6.3.32, MySQL Cluster NDB
7.0.13, and MySQL Cluster NDB 7.1.2, it is possible to use
this parameter to set the scheduling policy and priority of
heartbeat threads for management and API nodes.
The syntax for setting this parameter is shown here:
HeartbeatThreadPriority = policy [, priority ]
policy :
{FIFO | RR}
When setting this parameter, you must specify a policy. This
is one of FIFO (first in, first in) or
RR (round robin). This followed
optionally by the priority (an integer).
MaxScanBatchSize
The batch size is the size of each batch sent from each data
node. Most scans are performed in parallel to protect the
MySQL Server from receiving too much data from many nodes in
parallel; this parameter sets a limit to the total batch
size over all nodes.
The default value of this parameter is set to 256KB. Its
maximum size is 16MB.
TotalSendBufferMemory
This parameter is available beginning with MySQL Cluster NDB
6.4.0. It is used to determine the total amount of memory to
allocate on this node for shared send buffer memory among
all configured transporters.
If this parameter is set, its minimum allowed value is 256K;
the maxmimum is 4294967039. For more detailed information
about the behavior and use of
TotalSendBufferMemory and configuring
send buffer memory parameters in MySQL Cluster NDB 6.4.0 and
later, see
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”.
AutoReconnect
This parameter is available beginning with MySQL Cluster NDB
6.3.26 and MySQL Cluster NDB 7.0.7. By default, its value is
false , which forces disconnected API
nodes (including MySQL Servers acting as SQL nodes) the use
a new connection to the cluster rather than attempting to
re-use an existing one, which can cause problems when using
dynamically-allocated node IDs. (Bug#45921)
You can obtain some information from a MySQL server running as a
Cluster SQL node using SHOW
STATUS in the mysql client, as
shown here:
mysql> SHOW STATUS LIKE 'ndb%';
+-----------------------------+---------------+
| Variable_name | Value |
+-----------------------------+---------------+
| Ndb_cluster_node_id | 5 |
| Ndb_config_from_host | 192.168.0.112 |
| Ndb_config_from_port | 1186 |
| Ndb_number_of_storage_nodes | 4 |
+-----------------------------+---------------+
4 rows in set (0.02 sec)
For information about these Cluster system status variables, see
Section 5.1.7, “Server Status Variables”.
Note
To add new SQL or API nodes to the configuration of a running
MySQL Cluster, it is necessary to perform a rolling restart of
all cluster nodes after adding new [mysqld]
or [api] sections to the
config.ini file (or files, if you are
using more than one management server). This must be done
before the new SQL or API nodes can connect to the cluster.
It is not necessary to perform any
restart of the cluster if new SQL or API nodes can employ
previously unused API slots in the cluster configuration to
connect to the cluster.
17.3.2.8. MySQL Cluster TCP/IP Connections
TCP/IP is the default transport mechanism for all connections
between nodes in a MySQL Cluster. Normally it is not necessary
to define TCP/IP connections; MySQL Cluster automatically sets
up such connections for all data nodes, management nodes, and
SQL or API nodes.
To override the default connection parameters, it is necessary
to define a connection using one or more
[tcp] sections in the
config.ini file. Each
[tcp] section explicitly defines a TCP/IP
connection between two MySQL Cluster nodes, and must contain at
a minimum the parameters NodeId1 and
NodeId2 , as well as any connection parameters
to override.
It is also possible to change the default values for these
parameters by setting them in the [tcp
default] section.
Important
Any [tcp] sections in the
config.ini file should be listed
last, following all other sections in the
file. However, this is not required for a [tcp
default] section. This requirement is a known issue
with the way in which the config.ini file
is read by the MySQL Cluster management server.
Connection parameters which can be set in
[tcp] and [tcp default]
sections of the config.ini file are listed
here:
NodeId1 , NodeId2
To identify a connection between two nodes it is necessary
to provide their node IDs in the [tcp]
section of the configuration file. These are the same unique
Id values for each of these nodes as
described in Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
OverloadLimit
Beginning in MySQL Cluster NDB 6.4.0, this parameter can be
used to determine the amount of unsent data that must be
present in the send buffer before the connection is
considered overloaded. See
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”, for
more information.
SendBufferMemory
TCP transporters use a buffer to store all messages before
performing the send call to the operating system. When this
buffer reaches 64KB its contents are sent; these are also
sent when a round of messages have been executed. To handle
temporary overload situations it is also possible to define
a bigger send buffer.
Prior to MySQL Cluster NDB 7.0, this parameter determined a
fixed amount of memory allocated at startup for each
configured TCP connection. Beginning with MySQL Cluster NDB
6.4.0, memory is not dedicated to each transporter. Instead,
the value denotes the hard limit for how much memory (out of
the total available memory — that is,
TotalSendBufferMemory ) that may be used
by a single transporter. For more information about
configuring dynamic transporter send buffer memory
allocation in MySQL Cluster NDB 7.0 and later, see
Section 17.3.2.13, “Configuring MySQL Cluster Send Buffer Parameters”.
In MySQL Cluster NDB 6.4.3 and earlier, the default size of
the send buffer was 256 KB; in MySQL Cluster NDB 7.0.4 and
later, it is 2MB, which is the size recommended in most
situations. The minimum size is 64 KB; the theoretical
maximum is 4 GB.
SendSignalId
To be able to retrace a distributed message datagram, it is
necessary to identify each message. When this parameter is
set to Y , message IDs are transported
over the network. This feature is disabled by default in
production builds, and enabled in -debug
builds.
Checksum
This parameter is a boolean parameter (enabled by setting it
to Y or 1 , disabled by
setting it to N or 0 ).
It is disabled by default. When it is enabled, checksums for
all messages are calculated before they placed in the send
buffer. This feature ensures that messages are not corrupted
while waiting in the send buffer, or by the transport
mechanism.
PortNumber
(OBSOLETE)
This formerly specified the port number to be used for
listening for connections from other nodes. This parameter
should no longer be used.
ReceiveBufferMemory
Specifies the size of the buffer used when receiving data
from the TCP/IP socket.
In MySQL Cluster NDB 6.4.3 and earlier, the default value of
this parameter was 64 KB; beginning with MySQL Cluster NDB
7.0.4, 2MB is the default. The minimum possible value is
16KB; the theoretical maximum is 4GB.
17.3.2.9. MySQL Cluster TCP/IP Connections Using Direct Connections
Setting up a cluster using direct connections between data nodes
requires specifying explicitly the crossover IP addresses of the
data nodes so connected in the [tcp] section
of the cluster config.ini file.
In the following example, we envision a cluster with at least
four hosts, one each for a management server, an SQL node, and
two data nodes. The cluster as a whole resides on the
172.23.72.* subnet of a LAN. In addition to
the usual network connections, the two data nodes are connected
directly using a standard crossover cable, and communicate with
one another directly using IP addresses in the
1.1.0.* address range as shown:
# Management Server
[ndb_mgmd]
Id=1
HostName=172.23.72.20
# SQL Node
[mysqld]
Id=2
HostName=172.23.72.21
# Data Nodes
[ndbd]
Id=3
HostName=172.23.72.22
[ndbd]
Id=4
HostName=172.23.72.23
# TCP/IP Connections
[tcp]
NodeId1=3
NodeId2=4
HostName1=1.1.0.1
HostName2=1.1.0.2
The HostNameN
parameter, where N is an integer, is
used only when specifying direct TCP/IP connections.
The use of direct connections between data nodes can improve the
cluster's overall efficiency by allowing the data nodes to
bypass an Ethernet device such as a switch, hub, or router, thus
cutting down on the cluster's latency. It is important to note
that to take the best advantage of direct connections in this
fashion with more than two data nodes, you must have a direct
connection between each data node and every other data node in
the same node group.
17.3.2.10. MySQL Cluster Shared-Memory Connections
MySQL Cluster attempts to use the shared memory transporter and
configure it automatically where possible.
[shm] sections in the
config.ini file explicitly define
shared-memory connections between nodes in the cluster. When
explicitly defining shared memory as the connection method, it
is necessary to define at least NodeId1 ,
NodeId2 and ShmKey . All
other parameters have default values that should work well in
most cases.
Important
SHM functionality is considered experimental
only. It is not officially supported in any current
MySQL Cluster release, and testing results indicate that SHM
performance is not appreciably greater than when using TCP/IP
for the transporter.
For these reasons, you must determine for yourself or by using
our free resources (forums, mailing lists) whether SHM can be
made to work correctly in your specific case.
NodeId1 , NodeId2
To identify a connection between two nodes it is necessary
to provide node identifiers for each of them, as
NodeId1 and NodeId2 .
ShmKey
When setting up shared memory segments, a node ID, expressed
as an integer, is used to identify uniquely the shared
memory segment to use for the communication. There is no
default value.
ShmSize
Each SHM connection has a shared memory segment where
messages between nodes are placed by the sender and read by
the reader. The size of this segment is defined by
ShmSize . The default value is 1MB.
SendSignalId
To retrace the path of a distributed message, it is
necessary to provide each message with a unique identifier.
Setting this parameter to Y causes these
message IDs to be transported over the network as well. This
feature is disabled by default in production builds, and
enabled in -debug builds.
Checksum
This parameter is a boolean
(Y /N ) parameter which
is disabled by default. When it is enabled, checksums for
all messages are calculated before being placed in the send
buffer.
This feature prevents messages from being corrupted while
waiting in the send buffer. It also serves as a check
against data being corrupted during transport.
SigNum
When using the shared memory transporter, a process sends an
operating system signal to the other process when there is
new data available in the shared memory. Should that signal
conflict with with an existing signal, this parameter can be
used to change it. This is a possibility when using SHM due
to the fact that different operating systems use different
signal numbers.
The default value of SigNum is 0;
therefore, it must be set to avoid errors in the cluster log
when using the shared memory transporter. Typically, this
parameter is set to 10 in the [shm
default] section of the
config.ini file.
17.3.2.11. SCI Transport Connections in MySQL Cluster
[sci] sections in the
config.ini file explicitly define SCI
(Scalable Coherent Interface) connections between cluster nodes.
Using SCI transporters in MySQL Cluster is supported only when
the MySQL binaries are built using
--with-ndb-sci=/your/path/to/SCI .
The path should point to a directory
that contains at a minimum lib and
include directories containing SISCI
libraries and header files. (See
Section 17.3.5, “Using High-Speed Interconnects with MySQL Cluster” for more
information about SCI.)
In addition, SCI requires specialized hardware.
It is strongly recommended to use SCI Transporters only for
communication between ndbd processes. Note
also that using SCI Transporters means that the
ndbd processes never sleep. For this reason,
SCI Transporters should be used only on machines having at least
two CPUs dedicated for use by ndbd processes.
There should be at least one CPU per ndbd
process, with at least one CPU left in reserve to handle
operating system activities.
NodeId1 , NodeId2
To identify a connection between two nodes it is necessary
to provide node identifiers for each of them, as
NodeId1 and NodeId2 .
Host1SciId0
This identifies the SCI node ID on the first Cluster node
(identified by NodeId1 ).
Host1SciId1
It is possible to set up SCI Transporters for failover
between two SCI cards which then should use separate
networks between the nodes. This identifies the node ID and
the second SCI card to be used on the first node.
Host2SciId0
This identifies the SCI node ID on the second Cluster node
(identified by NodeId2 ).
Host2SciId1
When using two SCI cards to provide failover, this parameter
identifies the second SCI card to be used on the second
node.
SharedBufferSize
Each SCI transporter has a shared memory segment used for
communication between the two nodes. Setting the size of
this segment to the default value of 1MB should be
sufficient for most applications. Using a smaller value can
lead to problems when performing many parallel inserts; if
the shared buffer is too small, this can also result in a
crash of the ndbd process.
SendLimit
A small buffer in front of the SCI media stores messages
before transmitting them over the SCI network. By default,
this is set to 8KB. Our benchmarks show that performance is
best at 64KB but 16KB reaches within a few percent of this,
and there was little if any advantage to increasing it
beyond 8KB.
SendSignalId
To trace a distributed message it is necessary to identify
each message uniquely. When this parameter is set to
Y , message IDs are transported over the
network. This feature is disabled by default in production
builds, and enabled in -debug builds.
Checksum
This parameter is a boolean value, and is disabled by
default. When Checksum is enabled,
checksums are calculated for all messages before they are
placed in the send buffer. This feature prevents messages
from being corrupted while waiting in the send buffer. It
also serves as a check against data being corrupted during
transport.
17.3.2.12. Configuring MySQL Cluster Parameters for Local Checkpoints
The parameters discussed in
Logging
and Checkpointing and in
Data
Memory, Index Memory, and String Memory that are used to
configure local checkpoints for a MySQL Cluster do not exist in
isolation, but rather are very much interdepedent on each other.
In this section, we illustrate how these parameters —
including DataMemory ,
IndexMemory ,
NoOfDiskPagesToDiskAfterRestartTUP ,
NoOfDiskPagesToDiskAfterRestartACC , and
NoOfFragmentLogFiles — relate to one
another in a working Cluster.
Important
The parameters
NoOfDiskPagesToDiskAfterRestartTUP and
NoOfDiskPagesToDiskAfterRestartACC were
deprecated in MySQL 5.1.6. From MySQL 5.1.6 through 5.1.11,
disk writes during LCPs took place at the maximum speed
possible. Beginning with MySQL 5.1.12, the speed and
throughput for LCPs are controlled using the parameters
DiskSyncSize ,
DiskCheckpointSpeed , and
DiskCheckpointSpeedInRestart . See
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
In this example, we assume that our application performs the
following numbers of types of operations per hour:
50000 selects
15000 inserts
15000 updates
15000 deletes
We also make the following assumptions about the data used in
the application:
We are working with a single table having 40 columns.
Each column can hold up to 32 bytes of data.
A typical UPDATE run by the
application affects the values of 5 columns.
No NULL values are inserted by the
application.
A good starting point is to determine the amount of time that
should elapse between local checkpoints (LCPs). It is worth
noting that, in the event of a system restart, it takes 40-60
percent of this interval to execute the REDO log — for
example, if the time between LCPs is 5 minutes (300 seconds),
then it should take 2 to 3 minutes (120 to 180 seconds) for the
REDO log to be read.
The maximum amount of data per node can be assumed to be the
size of the DataMemory parameter. In this
example, we assume that this is 2 GB. The
NoOfDiskPagesToDiskAfterRestartTUP parameter
represents the amount of data to be checkpointed per unit time
— however, this parameter is actually expressed as the
number of 8K memory pages to be checkpointed per 100
milliseconds. 2 GB per 300 seconds is approximately 6.8 MB per
second, or 700 KB per 100 milliseconds, which works out to
roughly 85 pages per 100 milliseconds.
Similarly, we can calculate
NoOfDiskPagesToDiskAfterRestartACC in terms
of the time for local checkpoints and the amount of memory
required for indexes — that is, the
IndexMemory . Assuming that we allow 512 MB
for indexes, this works out to approximately 20 8-KB pages per
100 milliseconds for this parameter.
Next, we need to determine the number of REDO log files required
— that is, fragment log files — the corresponding
parameter being NoOfFragmentLogFiles . We need
to make sure that there are sufficient REDO log files for
keeping records for at least 3 local checkpoints (in MySQL
Cluster NDB 6.3.8 and later, we need only allow for 2 local
checkpoints). In a production setting, there are always
uncertainties — for instance, we cannot be sure that disks
always operate at top speed or with maximum throughput. For this
reason, it is best to err on the side of caution, so we double
our requirement and calculate a number of fragment log files
which should be enough to keep records covering 6 local
checkpoints (in MySQL Cluster NDB 6.3.8 and later, a number of
fragment log files accommodating 4 local checkpoints should be
sufficient).
It is also important to remember that the disk also handles
writes to the REDO log, so if you find that the amount of data
being written to disk as determined by the values of
NoOfDiskPagesToDiskAfterRestartACC and
NoOfDiskPagesToDiskAfterRestartTUP is
approaching the amount of disk bandwidth available, you may wish
to increase the time between local checkpoints.
Given 5 minutes (300 seconds) per local checkpoint, this means
that we need to support writing log records at maximum speed for
6 * 300 = 1800 seconds (MySQL Cluster NDB 6.3.8 and
later: 4 * 300 = 1200 seconds). The size of a REDO
log record is 72 bytes plus 4 bytes per updated column value
plus the maximum size of the updated column, and there is one
REDO log record for each table record updated in a transaction,
on each node where the data reside. Using the numbers of
operations set out previously in this section, we derive the
following:
50000 select operations per hour yields 0 log records (and
thus 0 bytes), since SELECT
statements are not recorded in the REDO log.
15000 DELETE statements per
hour is approximately 5 delete operations per second. (Since
we wish to be conservative in our estimate, we round up here
and in the following calculations.) No columns are updated
by deletes, so these statements consume only 5 operations *
72 bytes per operation = 360 bytes per second.
15000 UPDATE statements per
hour is roughly the same as 5 updates per second. Each
update uses 72 bytes, plus 4 bytes per column * 5 columns
updated, plus 32 bytes per column * 5 columns — this
works out to 72 + 20 + 160 = 252 bytes per operation, and
multiplying this by 5 operation per second yields 1260 bytes
per second.
15000 INSERT statements per
hour is equivalent to 5 insert operations per second. Each
insert requires REDO log space of 72 bytes, plus 4 bytes per
record * 40 columns, plus 32 bytes per column * 40 columns,
which is 72 + 160 + 1280 = 1512 bytes per operation. This
times 5 operations per second yields 7560 bytes per second.
So the total number of REDO log bytes being written per second
is approximately 0 + 360 + 1260 + 7560 = 9180 bytes. Multiplied
by 1800 seconds, this yields 16524000 bytes required for REDO
logging, or approximately 15.75 MB. The unit used for
NoOfFragmentLogFiles represents a set of 4
16-MB log files — that is, 64 MB. Thus, the minimum value
(3) for this parameter is sufficient for the scenario envisioned
in this example, since 3 times 64 = 192 MB, or about 12 times
what is required; the default value of 8 (or 512 MB) is more
than ample in this case.
17.3.2.13. Configuring MySQL Cluster Send Buffer Parameters
Formerly, the NDB kernel used a send buffer whose size was fixed
at 2MB for every node in the cluster, which was allocated when
the node started. Because the size of this buffer could not be
changed after the cluster was started, it was necessary to make
it large enough in advance to accomodate the maximum possible
load on any transporter socket. However, this was an inefficient
use of memory, since much of it often went unused, and could
result in large amounts of resources being wasted when scaling
up to many API nodes.
Beginning with MySQL Cluster NDB 7.0, this problem is solved by
employing a unified send buffer whose memory is allocated
dynamically from a pool shared by all transporters. This means
that the size of the send buffer can be adjusted as necessary.
Configuration of the unified send buffer can accomplished by
setting the following parameters:
TotalSendBufferMemory .
This parameter can be set for all types of MySQL Cluster
nodes — that is, it can be set in the
[ndbd] , [mgm] , and
[api] (or [mysql] )
sections of the config.ini file. It
represents the total amount of memory (in bytes) to be
allocated by each node for which it is set for use among
all configured transporters. If set, its minimum is 256K;
the maximum is 4294967039.
In order to be backward-compatible with existing
configurations, this parameter takes as its default value
the sum of the maximum send buffer sizes of all configured
transporters, plus an additional 32KB (one page) per
transporter. The maximum depends on the type of transporter,
as shown in the following table:
This allows existing configurations to function in close to
the same way as they did with MySQL Cluster NDB 6.3 and
earlier, with the same amount of memory and send buffer
space available to each transporter. However, memory that is
unused by one transporter is not available to other
transporters.
OverloadLimit .
This parameter is used in the
config.ini file
[tcp] section, and denotes the amount
of unsent data (in bytes) that must be present in the send
buffer before the connection is considered overloaded.
When such an overload condition occurs, transactions that
affect the overloaded connection fail with NDB API Error
1218 (Send Buffers overloaded in NDB
kernel) until the overload status passes. The
default value is 0, in which case the effective overload
limit is calculated as SendBufferMemory *
0.8 for a given connection. The maximum value
for this parameter is 4G.
SendBufferMemory .
In MySQL Cluster NDB 6.3 and earlier, this TCP
configuration parameter represented the amount of memory
allocated at startup for each configured TCP connection.
Beginning with MySQL Cluster NDB 7.0, this value denotes a
hard limit for the amount of memory that may be used by a
single transporter out of the entire pool specified by
TotalSendBufferMemory . However, the sum
of SendBufferMemory for all configured
transporters may be greater than the
TotalSendBufferMemory that is set for a
given node. This is a way to save memory when many nodes
are in use, as long as the maximum amount of memory is
never required by all transporters at the same time.
17.3.3. Overview of MySQL Cluster Configuration Parameters
The next four sections provide summary tables of MySQL Cluster
configuration parameters used in the
config.ini file to govern the cluster's
functioning. Each table lists the parameters for one of the
Cluster node process types (ndbd,
ndb_mgmd, and mysqld), and
includes the parameter's type as well as its default, mimimum, and
maximum values as applicable.
It is also stated what type of restart is required (node restart
or system restart) — and whether the restart must be done
with --initial — to change the value of a
given configuration parameter.
When performing a node restart or an initial node restart, all of
the cluster's data nodes must be restarted in turn (also referred
to as a rolling restart). It is possible to
update cluster configuration parameters marked as
node online — that is, without shutting
down the cluster — in this fashion. An initial node restart
requires restarting each ndbd process with the
--initial option.
A system restart requires a complete shutdown and restart of the
entire cluster. An initial system restart requires taking a backup
of the cluster, wiping the cluster file system after shutdown, and
then restoring from the backup following the restart.
In any cluster restart, all of the cluster's management servers
must be restarted in order for them to read the updated
configuration parameter values.
Important
Values for numeric cluster parameters can generally be increased
without any problems, although it is advisable to do so
progressively, making such adjustments in relatively small
increments. Many of these can be increased online, using a
rolling restart.
However, decreasing the values of such parameters —
whether this is done via a node restart, node initial restart,
or even a complete system restart of the cluster — is not
to be undertaken lightly; it is recommended that you do so only
after careful planning and testing. This is especially true with
regard to those parameters that relate to memory usage and disk
space, such as MaxNoOfTables ,
MaxNoOfOrderedIndexes , and
MaxNoOfUniqueHashIndexes . In addition, it is
the generally the case that configuration parameters relating to
memory and disk usage can be raised using a simple node restart,
but they require an initial node restart to be lowered.
Because some of these parameters can be used for configuring more
than one type of cluster node, they may appear in more than one of
the tables.
Note
4294967039 — which often appears as a
maximum value in these tables — is defined in the
NDBCLUSTER sources as
MAX_INT_RNIL and is equal to
0xFFFFFEFF , or
232 –
28 – 1 .
17.3.3.1. MySQL Cluster Data Node Configuration Parameters
The following table provides information about parameters used
in the [ndbd] or [ndbd
default] sections of a config.ini
file for configuring MySQL Cluster data nodes. For detailed
descriptions and other additional information about each of
these parameters, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
Beginning with MySQL Cluster NDB 6.4.0, these parameters also
apply to ndbmtd, which is a multi-threaded
version of ndbd. For more information, see
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”.
Table 17.1. Data Node Configuration Parameters Note
To add new data nodes to a MySQL Cluster, it is necessary to
shut down the cluster completely, update the
config.ini file, and then restart the
cluster (that is, you must perform a system restart). All data
node processes must be started with the
--initial option.
Beginning in MySQL Cluster NDB 7.0, it is possible to add new
data node groups to a running cluster online. For more
information, see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
17.3.3.2. MySQL Cluster Management Node Configuration Parameters
The following table provides information about parameters used
in the [ndb_mgmd] or [mgm]
sections of a config.ini file for
configuring MySQL Cluster management nodes. For detailed
descriptions and other additional information about each of
these parameters, see
Section 17.3.2.5, “Defining a MySQL Cluster Management Server”.
Table 17.2. Management Node Configuration Parameters 17.3.3.3. MySQL Cluster SQL Node and API Node Configuration Parameters
The following table provides information about parameters used
in the [SQL] and [api]
sections of a config.ini file for
configuring MySQL Cluster SQL nodes and API nodes. For detailed
descriptions and other additional information about each of
these parameters, see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
Table 17.3. API Node Configuration Parameters Note
To add new SQL or API nodes to the configuration of a running
MySQL Cluster, it is necessary to perform a rolling restart of
all cluster nodes after adding new [mysqld]
or [api] sections to the
config.ini file (or files, if you are
using more than one management server). This must be done
before the new SQL or API nodes can connect to the cluster.
It is not necessary to perform any
restart of the cluster if new SQL or API nodes can employ
previously unused API slots in the cluster configuration to
connect to the cluster.
17.3.3.4. Other MySQL Cluster Configuration Parameters
The following tables provide information about parameters used
in the [computer] , [tcp] ,
[shm] , and [sci] sections
of a config.ini file for configuring MySQL
Cluster management nodes. For detailed descriptions and other
additional information about individual parameters, see
Section 17.3.2.8, “MySQL Cluster TCP/IP Connections”,
Section 17.3.2.10, “MySQL Cluster Shared-Memory Connections”, or
Section 17.3.2.11, “SCI Transport Connections in MySQL Cluster”, as appropriate.
Table 17.4. COMPUTER Configuration Parameters Name | Type/Units | Default | Min Value | Max Value | Restart Type |
---|
HostName | name or IP | | | | S | Id | string | | | | IN |
Table 17.5. TCP Configuration Parameters Name | Type/Units | Default | Min Value | Max Value | Restart Type |
---|
Checksum | | false | | | N | Group | unsigned | 55 | | 200 | N | NodeId1 | | | | | N | NodeId2 | | | | | N | NodeIdServer | | | | | N | OverloadLimit | bytes | | | 4G | N | PortNumber | unsigned | | | 64K | N | Proxy | | | | | N | ReceiveBufferMemory | bytes | 64K | 16K | 4G | N | SendBufferMemory | unsigned | 256K | 64K | 4G | N | SendSignalId | | true (debug builds: false) | | | N | TCP_MAXSEG_SIZE | unsigned | | | 2G | N | TCP_RCV_BUF_SIZE | unsigned | 70080 | 1 | 2G | N | TCP_SND_BUF_SIZE | unsigned | 71540 | 1 | 2G | N | TcpBind_INADDR_ANY | | false | | | N |
Table 17.6. SHM Configuration Parameters Name | Type/Units | Default | Min Value | Max Value | Restart Type |
---|
Checksum | | true | | | N | Group | unsigned | 35 | | 200 | N | NodeId1 | | | | | N | NodeId2 | | | | | N | NodeIdServer | | | | | N | OverloadLimit | bytes | | | 4G | N | PortNumber | unsigned | | | 64K | N | SendSignalId | | false | | | N | ShmKey | unsigned | | | 4G | N | ShmSize | bytes | 1M | 64K | 4G | N | Signum | unsigned | | | 4G | N |
Table 17.7. SCI Configuration Parameters 17.3.4. MySQL Server Options and Variables for MySQL Cluster
This section provides information about MySQL server options,
server and status variables that are specific to MySQL Cluster.
For general information on using these, and for other options and
variables not specific to MySQL Cluster, see
Section 5.1, “The MySQL Server”.
For MySQL Cluster configuration parameters used in the cluster
confiuration file (usually named config.ini ),
see Section 17.3, “MySQL Cluster Configuration”.
17.3.4.1. MySQL Cluster Server Option and Variable Reference
The following table provides a list of the command-line options,
server and status variables applicable within
mysqld when it is running as an SQL node in a
MySQL Cluster. For a table showing all
command-line options, server and status variables available for
use with mysqld, see
Section 5.1.1, “Server Option and Variable Reference”.
Table 17.8. Command Options for MySQL Cluster 17.3.4.2. mysqld Command Options for MySQL Cluster
This section provides descriptions of mysqld
server options relating to MySQL Cluster. For information about
mysqld options not specific to MySQL Cluster,
and for general information about the use of options with
mysqld, see Section 5.1.2, “Server Command Options”.
For information about command-line options used with other MySQL
Cluster processes (ndbd,
ndb_mgmd, and ndb_mgm),
see Section 17.4.23, “Options Common to MySQL Cluster Programs”. For
information about command-line options used with
NDB utility programs (such as
ndb_desc, ndb_size.pl, and
ndb_show_tables), see
Section 17.4, “MySQL Cluster Programs”.
--ndb-batch-size=#
This sets the size in bytes that is used for NDB transaction
batches.
--ndb-cluster-connection-pool=#
By setting this option to a value greater than 1 (the
default), a mysqld process can use
multiple connections to the cluster, effectively mimicking
several SQL nodes. Each connection requires its own
[api] or [mysqld]
section in the cluster configuration
(config.ini ) file, and counts against
the maximum number of API connections supported by the
cluster.
Suppose that you have 2 cluster host computers, each running
an SQL node whose mysqld process was
started with
--ndb-cluster-connection-pool=4 ; this means
that the cluster must have 8 API slots available for these
connections (instead of 2). All of these connections are set
up when the SQL node connects to the cluster, and are
allocated to threads in a round-robin fashion.
This option is useful only when running
mysqld on host machines having multiple
CPUs, multiple cores, or both. For best results, the value
should be smaller than the total number of cores available
on the host machine. Setting it to a value greater than this
is likely to degrade performance severely.
Important
Because each SQL node using connection pooling occupies
multiple API node slots — each slot having its own
node ID in the cluster — you must
not use a node ID as part of the
cluster connect string when starting any
mysqld process that employs connection
pooling.
Setting a node ID in the connect string when using the
--ndb-cluster-connection-pool option
causes node ID allocation errors when the SQL node
attempts to connect to the cluster.
This option was introduced in MySQL Cluster NDB 6.2.2.
Beginning with MySQL Cluster NDB 6.2.16 and MySQL Cluster
NDB 6.3.13, the value used for this option is available as a
global status variable (Bug#35573).
--ndb-connectstring=connect_string
When using the NDBCLUSTER
storage engine, this option specifies the management server
that distributes cluster configuration data. See
Section 17.3.2.3, “The MySQL Cluster Connectstring”, for syntax.
--ndbcluster
The NDBCLUSTER storage engine
is necessary for using MySQL Cluster. If a
mysqld binary includes support for the
NDBCLUSTER storage engine, the
engine is disabled by default. Use the
--ndbcluster option to enable
it. Use --skip-ndbcluster to explicitly
disable the engine.
--ndb-nodeid=#
Set this MySQL server's node ID in a MySQL Cluster.
This can be used instead of specifying the node ID as part
of the connectstring or in the
config.ini file, or allowing the
cluster to determine an arbitrary node ID. If you use this
option, then --ndb-nodeid must be specified
before
--ndb-connectstring . If
--ndb-nodeid is used
and a node ID is specified in the
connectstring, then the MySQL server will not be able to
connect to the cluster. In addition, if
--nodeid is used, then either a matching
node ID must be found in a [mysqld] or
[api] section of
config.ini , or there must be an
“open” [mysqld] or
[api] section in the file (that is, a
section without an Id parameter
specified).
Regardless of how the node ID is determined, its is shown as
the value of the global status variable
Ndb_cluster_node_id in the output of
SHOW STATUS , and as
cluster_node_id in the
connection row of the output of
SHOW ENGINE
NDBCLUSTER STATUS .
For more information about node IDs for MySQL Cluster SQL
nodes, see Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
--skip-ndbcluster
Disable the NDBCLUSTER storage
engine. This is the default for binaries that were built
with NDBCLUSTER storage engine
support; the server allocates memory and other resources for
this storage engine only if the
--ndbcluster option is given
explicitly. See Section 17.3.1, “Quick Test Setup of MySQL Cluster”, for
an example.
17.3.4.3. MySQL Cluster System Variables
This section provides detailed information about MySQL server
system variables that are specific to MySQL Cluster and the
NDB storage engine. For system
variables not specific to MySQL Cluster, see
Section 5.1.4, “Server System Variables”. For general
information on using system variables, see
Section 5.1.6, “Using System Variables”.
have_ndbcluster
YES if mysqld supports
NDBCLUSTER tables.
DISABLED if
--skip-ndbcluster is used.
multi_range_count
The maximum number of ranges to send to a table handler at
once during range selects. The default value is 256. Sending
multiple ranges to a handler at once can improve the
performance of certain selects dramatically. This is
especially true for the
NDBCLUSTER table handler, which
needs to send the range requests to all nodes. Sending a
batch of those requests at once reduces communication costs
significantly.
This variable is deprecated in MySQL 5.1, and is no longer
supported in MySQL 5.4, in which arbitrarily long lists of
ranges can be processed.
ndb_autoincrement_prefetch_sz
Determines the probability of gaps in an autoincremented
column. Set it to 1 to minimize this.
Setting it to a high value for optimization — makes
inserts faster, but decreases the likelihood that
consecutive autoincrement numbers will be used in a batch of
inserts. Default value: 32 . Minimum
value: 1 .
Beginning with MySQL Cluster NDB 6.2.10, MySQL Cluster NDB
6.3.7, and MySQL 5.1.23, this variable affects the number of
AUTO_INCREMENT IDs that are fetched
between statements only. Within a statement, at least 32 IDs
are now obtained at a time. The default value for
ndb_autoincrement_prefetch_sz
is now 1 , to increase the speed of
statements inserting single rows. (Bug#31956)
Beginning with MySQL Cluster NDB 6.3.31 and MySQL CLuster
NDB 7.0.11, the maximum value for
ndb_autoincrement_prefetch_sz is
increased, from 256 to 65536. (Bug#50621)
ndb_cache_check_time
The number of milliseconds that elapse between checks of
MySQL Cluster SQL nodes by the MySQL query cache. Setting
this to 0 (the default and minimum value) means that the
query cache checks for validation on every query.
The recommended maximum value for this variable is 1000,
which means that the check is performed once per second. A
larger value means that the check is performed and possibly
invalidated due to updates on different SQL nodes less
often. It is generally not desirable to set this to a value
greater than 2000.
ndb_extra_logging
This variable can be used to enable recording in the MySQL
error log of information specific to the
NDB storage engine. It is
normally of interest only when debugging
NDB storage engine code.
The default value is 0, which means that the only
NDB -specific information
written to the MySQL error log relates to transaction
handling. If the value is greater than 0 but less than 10,
NDB table schema and connection
events are also logged, as well as whether or not conflict
resolution is in use, and other
NDB errors and information. If
the value is set to 10 or more, information about
NDB internals, such as the
progress of data distribution among cluster nodes, is also
written to the MySQL error log.
This variable was added in MySQL 5.1.6.
ndb_force_send
Forces sending of buffers to
NDB immediately, without
waiting for other threads. Defaults to
ON .
ndb_index_stat_cache_entries
Sets the granularity of the statistics by determining the
number of starting and ending keys to store in the
statistics memory cache. Zero means no caching takes place;
in this case, the data nodes are always queried directly.
Default value: 32 .
ndb_index_stat_enable
Use NDB index statistics in
query optimization. Defaults to ON .
ndb_index_stat_update_freq
How often to query data nodes instead of the statistics
cache. For example, a value of 20 (the
default) means to direct every
20th query to the data nodes.
Note
If
ndb_index_stat_cache_entries
is 0 , then setting this variable has no
effect; in this case, every query is sent directly to the
data nodes.
ndb_optimized_node_selection
Prior to MySQL Cluster NDB 6.3.4.
Causes an SQL node to use the “closest” data
node as transaction coordinator. Enabled by default. Set
to 0 or OFF to
disable, in which case the SQL node uses each data node in
the cluster in succession. When this option is disabled
each SQL thread attempts to use a given data node 8 times
before proceeding to the next one.
Beginning with MySQL Cluster NDB 6.3.4.
There are two forms of optimized node selection:
The SQL node uses promixity
to determine the transaction coordinator; that is,
the “closest” data node to the SQL node
is chosen as the transaction coordinator. For this
purpose, a data node having a shared memory
connection with the SQL node is considered to be
“closest” to the SQL node; the next
closest (in order of decreasing proximity) are: TCP
connection to localhost ; SCI
connection; TCP connection from a host other than
localhost .
The SQL thread uses distribution
awareness to select the data node. That
is, the data node housing the cluster partition
accessed by the first statement of a given
transaction is used as the transaction coordinator
for the entire transaction. (This is effective only
if the first statement of the transaction accesses
no more than one cluster partition.)
This option takes one of the integer values
0 , 1 ,
2 , or 3 .
3 is the default. These values affect
node selection as follows:
0 : Node selection is not
optimized. Each data node is employed as the
transaction coordinator 8 times before the SQL
thread proceeds to the next data node. (This is the
same “round-robin” behavior as caused
by setting this option to 0 or
OFF in previous versions of MySQL
Cluster.)
1 : Proximity to the SQL node is
used to determine the transaction coordinator. (This
is the same behavior as caused by setting this
option to 1 or
ON in previous MySQL versions.)
2 : Distribution awareness is used
to select the transaction coordinator. However, if
the first statement of the transaction accesses more
than one cluster partition, the SQL node reverts to
the round-robin behavior seen when this option is
set to 0 .
3 : If distribution awareness can
be employed to determine the transaction
coordinator, then it is used; otherwise proximity is
used to select the transaction coordinator. (This is
the default behavior in MySQL Cluster NDB 6.3.4 and
later.)
Important
Beginning with MySQL Cluster NDB 6.3.4, it is no longer
possible to set
--ndb_optimized_node_selection to
ON or OFF ;
attempting to do so causes mysqld to
abort with an error.
ndb_report_thresh_binlog_epoch_slip
This is a threshold on the number of epochs to be behind
before reporting binlog status. For example, a value of
3 (the default) means that if the
difference between which epoch has been received from the
storage nodes and which epoch has been applied to the binlog
is 3 or more, a status message will be sent to the cluster
log.
ndb_report_thresh_binlog_mem_usage
This is a threshold on the percentage of free memory
remaining before reporting binlog status. For example, a
value of 10 (the default) means that if
the amount of available memory for receiving binlog data
from the data nodes falls below 10%, a status message will
be sent to the cluster log.
ndb_use_copying_alter_table
Forces NDB to use copying of
tables in the event of problems with online
ALTER TABLE operations. The
default value is OFF .
This variable was added in MySQL 5.1.12.
ndb_use_exact_count
Forces NDB to use a count of
records during SELECT COUNT(*) query
planning to speed up this type of query. The default value
is ON . For faster queries overall,
disable this feature by setting the value of
ndb_use_exact_count to
OFF .
ndb_use_transactions
You can disable NDB transaction
support by setting this variable's values to
OFF (not recommended). The default is
ON .
Note
The setting for this variable was not honored in MySQL
Cluster NDB 6.4.3 and MySQL Cluster NDB 7.0.4. (Bug#43236)
ndb_wait_connected
This variable shows the period of time that the MySQL server
waits for connections to MySQL Cluster management and data
nodes to be established before accepting MySQL client
connections. The time is specified in seconds. The default
value is 0 .
ndb_wait_setup
This variable shows the period of time that the MySQL server
waits for the NDB storage
engine to complete setup before timing out and treating
NDB as unavailable. The time is
specified in seconds. The default value is
15 .
17.3.4.4. MySQL Cluster Status Variables
This section provides detailed information about MySQL server
status variables that relate to MySQL Cluster and the
NDB storage engine. For status
variables not specific to MySQL Cluster, and for general
information on using status variables, see
Section 5.1.7, “Server Status Variables”.
Handler_discover
The MySQL server can ask the
NDBCLUSTER storage engine if it
knows about a table with a given name. This is called
discovery.
Handler_discover indicates
the number of times that tables have been discovered via
this mechanism.
Ndb_cluster_node_id
If the server is acting as a MySQL Cluster node, then the
value of this variable its node ID in the cluster.
If the server is not part of a MySQL Cluster, then the value
of this variable is 0.
Ndb_config_from_host
If the server is part of a MySQL Cluster, the value of this
variable is the host name or IP address of the Cluster
management server from which it gets its configuration data.
If the server is not part of a MySQL Cluster, then the value
of this variable is an empty string.
Prior to MySQL 5.1.12, this variable was named
Ndb_connected_host .
Ndb_config_from_port
If the server is part of a MySQL Cluster, the value of this
variable is the number of the port through which it is
connected to the Cluster management server from which it
gets its configuration data.
If the server is not part of a MySQL Cluster, then the value
of this variable is 0.
Prior to MySQL 5.1.12, this variable was named
Ndb_connected_port .
Ndb_execute_count
Provides the number of round trips to the
NDB kernel made by operations.
Added in MySQL Cluster NDB 6.3.6.
Ndb_number_of_data_nodes
If the server is part of a MySQL Cluster, the value of this
variable is the number of data nodes in the cluster.
If the server is not part of a MySQL Cluster, then the value
of this variable is 0.
Prior to MySQL 5.1.12, this variable was named
Ndb_number_of_storage_nodes .
Slave_heartbeat_period
Shows the replication heartbeat interval (in seconds) on a
replication slave.
This variable was added in MySQL Cluster NDB 6.3.4.
Slave_received_heartbeats
This counter increments with each replication heartbeat
received by a replication slave since the last time that the
slave was restarted or reset, or a
CHANGE MASTER TO statement
was issued.
This variable was added in MySQL Cluster NDB 6.3.4.
Ndb_pruned_scan_count
This variable holds a count of the number of scans executed
by NDBCLUSTER since the MySQL
Cluster was last started where
NDBCLUSTER was able to use
partition pruning.
Using this variable together with
Ndb_scan_count can be
helpful in schema design to maximize the ability of the
server to prune scans to a single table partition, thereby
involving only a single data node.
This variable was added in MySQL Cluster NDB 6.3.25 and
MySQL Cluster NDB 7.0.5.
Ndb_scan_count
This variable holds a count of the total number of scans
executed by NDBCLUSTER since
the MySQL Cluster was last started.
This variable was added in MySQL Cluster NDB 6.3.25 and
MySQL Cluster NDB 7.0.5.
17.3.5. Using High-Speed Interconnects with MySQL Cluster
Even before design of NDBCLUSTER
began in 1996, it was evident that one of the major problems to be
encountered in building parallel databases would be communication
between the nodes in the network. For this reason,
NDBCLUSTER was designed from the very
beginning to allow for the use of a number of different data
transport mechanisms. In this Manual, we use the term
transporter for these.
The MySQL Cluster codebase provides for four different
transporters:
TCP/IP using 100 Mbps or gigabit
Ethernet, as discussed in
Section 17.3.2.8, “MySQL Cluster TCP/IP Connections”.
Direct (machine-to-machine) TCP/IP;
although this transporter uses the same TCP/IP protocol as
mentioned in the previous item, it requires setting up the
hardware differently and is configured differently as well.
For this reason, it is considered a separate transport
mechanism for MySQL Cluster. See
Section 17.3.2.9, “MySQL Cluster TCP/IP Connections Using Direct Connections”, for
details.
Shared memory (SHM). For more information
about SHM, see Section 17.3.2.10, “MySQL Cluster Shared-Memory Connections”.
Note
SHM is considered experimental only, and is not officially
supported.
Scalable Coherent Interface (SCI), as
described in the next section of this chapter,
Section 17.3.2.11, “SCI Transport Connections in MySQL Cluster”.
Most users today employ TCP/IP over Ethernet because it is
ubiquitous. TCP/IP is also by far the best-tested transporter for
use with MySQL Cluster.
We are working to make sure that communication with the
ndbd process is made in “chunks”
that are as large as possible because this benefits all types of
data transmission.
For users who desire it, it is also possible to use cluster
interconnects to enhance performance even further. There are two
ways to achieve this: Either a custom transporter can be designed
to handle this case, or you can use socket implementations that
bypass the TCP/IP stack to one extent or another. We have
experimented with both of these techniques using the SCI (Scalable
Coherent Interface) technology developed by
Dolphin Interconnect
Solutions.
17.3.5.1. Configuring MySQL Cluster to use SCI Sockets
It is possible employing Scalable Coherent Interface (SCI)
technology to achieve a significant increase in connection
speeds and throughput between MySQL Cluster data and SQL nodes.
To use SCI, it is necessary to obtain and install Dolphin SCI
network cards and to use the drivers and other software supplied
by Dolphin. You can get information on obtaining these, from
Dolphin Interconnect
Solutions. SCI SuperSocket or SCI Transporter support is
available for 32-bit and 64-bit Linux, Solaris, Windows, and
other platforms. See the Dolphin documentation referenced later
in this section for more detailed information regarding
platforms supported for SCI.
Note
Prior to MySQL 5.1.20, there were issues with building MySQL
Cluster with SCI support (see Bug#25470), but these have been
resolved due to work contributed by Dolphin. SCI Sockets are
now correctly supported for MySQL Cluster hosts running recent
versions of Linux using the -max builds,
and versions of MySQL Cluster with SCI Transporter support can
be built using either of
compile-amd64-max-sci or
compile-pentium64-max-sci. Both of these
build scripts can be found in the BUILD
directory of the MySQL Cluster source trees; it should not be
difficult to adapt them for other platforms. Generally, all
that is necessary is to compile MySQL Cluster with SCI
Transporter support is to configure the MySQL Cluster build
using --with-ndb-sci=/opt/DIS .
Once you have acquired the required Dolphin hardware and
software, you can obtain detailed information on how to adapt a
MySQL Cluster configured for normal TCP/IP communication to use
SCI from the Dolphin Express for MySQL Installation
and Reference Guide, available for download at
http://docsrva.mysql.com/public/DIS_install_guide_book.pdf
(PDF file, 94 pages, 753 KB). This document provides
instructions for installing the SCI hardware and software, as
well as information concerning network topology and
configuration.
17.3.5.2. MySQL Cluster Interconnects and Performance
The ndbd process has a number of simple
constructs which are used to access the data in a MySQL Cluster.
We have created a very simple benchmark to check the performance
of each of these and the effects which various interconnects
have on their performance.
There are four access methods:
Primary key access.
This is access of a record through its primary key. In the
simplest case, only one record is accessed at a time,
which means that the full cost of setting up a number of
TCP/IP messages and a number of costs for context
switching are borne by this single request. In the case
where multiple primary key accesses are sent in one batch,
those accesses share the cost of setting up the necessary
TCP/IP messages and context switches. If the TCP/IP
messages are for different destinations, additional TCP/IP
messages need to be set up.
Unique key access.
Unique key accesses are similar to primary key accesses,
except that a unique key access is executed as a read on
an index table followed by a primary key access on the
table. However, only one request is sent from the MySQL
Server, and the read of the index table is handled by
ndbd. Such requests also benefit from
batching.
Full table scan.
When no indexes exist for a lookup on a table, a full
table scan is performed. This is sent as a single request
to the ndbd process, which then divides
the table scan into a set of parallel scans on all cluster
ndbd processes. In future versions of
MySQL Cluster, an SQL node will be able to filter some of
these scans.
Range scan using ordered
index
When an ordered index is used, it performs a scan in the
same manner as the full table scan, except that it scans
only those records which are in the range used by the query
transmitted by the MySQL server (SQL node). All partitions
are scanned in parallel when all bound index attributes
include all attributes in the partitioning key.
With benchmarks developed internally by MySQL for testing simple
and batched primary and unique key accesses, we have found that
using SCI sockets improves performance by approximately 100%
over TCP/IP, except in rare instances when communication
performance is not an issue. This can occur when scan filters
make up most of processing time or when very large batches of
primary key accesses are achieved. In that case, the CPU
processing in the ndbd processes becomes a
fairly large part of the overhead.
Using the SCI transporter instead of SCI Sockets is only of
interest in communicating between ndbd
processes. Using the SCI transporter is also only of interest if
a CPU can be dedicated to the ndbd process
because the SCI transporter ensures that this process will never
go to sleep. It is also important to ensure that the
ndbd process priority is set in such a way
that the process does not lose priority due to running for an
extended period of time, as can be done by locking processes to
CPUs in Linux 2.6. If such a configuration is possible, the
ndbd process will benefit by 10–70% as
compared with using SCI sockets. (The larger figures will be
seen when performing updates and probably on parallel scan
operations as well.)
There are several other optimized socket implementations for
computer clusters, including Myrinet, Gigabit Ethernet,
Infiniband and the VIA interface. However, we have tested MySQL
Cluster so far only with SCI sockets. See
Section 17.3.5.1, “Configuring MySQL Cluster to use SCI Sockets”, for information on
how to set up SCI sockets using ordinary TCP/IP for MySQL
Cluster.
17.4. MySQL Cluster Programs
Using and managing a MySQL Cluster requires several specialized
programs, which we describe in this chapter. We discuss the purposes
of these programs in a MySQL Cluster, how to use the programs, and
what startup options are available for each of them.
These programs include the MySQL Cluster data, management, and SQL
node processes (ndbd, ndbmtd,
ndb_mgmd, and mysqld) and the
management client (ndb_mgm).
Other NDB utility, diagnostic, and example programs are included
with the MySQL Cluster distribution. These include
ndb_restore, ndb_show_tables,
and ndb_config. These programs are covered later
in this chapter.
The last two sections of this chapter contain tables of options
used, respectively, with mysqld and with the
various NDB programs.
17.4.1. MySQL Server Usage for MySQL Cluster
mysqld is the traditional MySQL server process.
To be used with MySQL Cluster, mysqld needs to
be built with support for the
NDBCLUSTER storage engine, as it is
in the precompiled binaries available from
http://dev.mysql.com/downloads/. If you build MySQL from
source, you must invoke configure with the
--with-ndbcluster option to enable NDB
Cluster storage engine support.
For information about other MySQL server options and variables
relevant to MySQL Cluster in addition to those discussed in this
section, see Section 17.3.4, “MySQL Server Options and Variables for MySQL Cluster”.
If the mysqld binary has been built with
Cluster support, the NDBCLUSTER
storage engine is still disabled by default. You can use either of
two possible options to enable this engine:
An easy way to verify that your server is running with the
NDBCLUSTER storage engine enabled is
to issue the SHOW ENGINES statement
in the MySQL Monitor (mysql). You should see
the value YES as the Support
value in the row for NDBCLUSTER . If
you see NO in this row or if there is no such
row displayed in the output, you are not running an
NDB -enabled version of MySQL. If you
see DISABLED in this row, you need to enable it
in either one of the two ways just described.
To read cluster configuration data, the MySQL server requires at a
minimum three pieces of information:
The MySQL server's own cluster node ID
The host name or IP address for the management server (MGM
node)
The number of the TCP/IP port on which it can connect to the
management server
Node IDs can be allocated dynamically, so it is not strictly
necessary to specify them explicitly.
The mysqld parameter
ndb-connectstring is used to specify the
connectstring either on the command line when starting
mysqld or in my.cnf . The
connectstring contains the host name or IP address where the
management server can be found, as well as the TCP/IP port it
uses.
In the following example, ndb_mgmd.mysql.com is
the host where the management server resides, and the management
server listens for cluster messages on port 1186:
shell> mysqld --ndbcluster --ndb-connectstring=ndb_mgmd.mysql.com:1186
See Section 17.3.2.3, “The MySQL Cluster Connectstring”, for more
information on connectstrings.
Given this information, the MySQL server will be a full
participant in the cluster. (We often refer to a
mysqld process running in this manner as an SQL
node.) It will be fully aware of all cluster data nodes as well as
their status, and will establish connections to all data nodes. In
this case, it is able to use any data node as a transaction
coordinator and to read and update node data.
You can see in the mysql client whether a MySQL
server is connected to the cluster using SHOW
PROCESSLIST . If the MySQL server is connected to the
cluster, and you have the PROCESS
privilege, then the first row of the output is as shown here:
mysql> SHOW PROCESSLIST \G
*************************** 1. row ***************************
Id: 1
User: system user
Host:
db:
Command: Daemon
Time: 1
State: Waiting for event from ndbcluster
Info: NULL
Important
To participate in a MySQL Cluster, the mysqld
process must be started with both the
options --ndbcluster and
--ndb-connectstring (or their equivalents in
my.cnf ). If mysqld is
started with only the
--ndbcluster option, or if it is
unable to contact the cluster, it is not possible to work with
NDB tables, nor is it
possible to create any new tables regardless of storage
engine. The latter restriction is a safety measure
intended to prevent the creation of tables having the same names
as NDB tables while the SQL node is
not connected to the cluster. If you wish to create tables using
a different storage engine while the mysqld
process is not participating in a MySQL Cluster, you must
restart the server without the
--ndbcluster option.
17.4.2. ndbd — The MySQL Cluster Data Node Daemon
ndbd is the process that is used to handle
all the data in tables using the NDB Cluster storage engine.
This is the process that empowers a data node to accomplish
distributed transaction handling, node recovery, checkpointing
to disk, online backup, and related tasks.
In a MySQL Cluster, a set of ndbd processes
cooperate in handling data. These processes can execute on the
same computer (host) or on different computers. The
correspondences between data nodes and Cluster hosts is
completely configurable.
The following table includes command options specific to the
MySQL Cluster data node program ndbd.
Additional descriptions follow the table. For options common
to all MySQL Cluster programs, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.9. ndbd Command Line
Options Format | Description | Introduction | Deprecated | Removed |
---|
--bind-address=name | Local bind address | 5.1.12 | | | --daemon | Start ndbd as daemon (default); override with --nodaemon | | | | --foreground | Run ndbd in foreground, provided for debugging purposes (implies --nodaemon) | | | | --initial | Perform initial start of ndbd, including cleaning the file system. Consult the documentation before using this option | | | | --initial-start | Perform partial initial start (requires --nowait-nodes) | 5.1.11 | | | --nodaemon | Do not start ndbd as daemon; provided for testing purposes | | | | --nostart | Don't start ndbd immediately; ndbd waits for command to start from ndb_mgmd | | | | --nowait-nodes=list | Do not wait for these data nodes to start (takes comma-separated list of node IDs). Also requires --ndb-nodeid to be used. | 5.1.9 | | |
Note
All of these options also apply to the multi-threaded
version of this program — ndbmtd,
which is available in MySQL Cluster NDB 7.0 — and you
may substitute “ndbmtd” for
“ndbd” wherever the latter
occurs in this section.
For options common to all
NDBCLUSTER programs, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
--bind-address
Causes ndbd to bind to a specific
network interface (host name or IP address). This option
has no default value.
This option was added in MySQL 5.1.12.
--daemon , -d
Instructs ndbd to execute as a daemon
process. This is the default behavior.
--nodaemon can be used to prevent the
process from running as a daemon.
This option has no effect when running
ndbd or ndbmtd on
Windows platforms.
--initial
Instructs ndbd to perform an initial
start. An initial start erases any files created for
recovery purposes by earlier instances of
ndbd. It also re-creates recovery log
files. Note that on some operating systems this process
can take a substantial amount of time.
An --initial start is to be used
only when starting the
ndbd process under very special
circumstances; this is because this option causes all
files to be removed from the Cluster file system and all
redo log files to be re-created. These circumstances are
listed here:
When performing a software upgrade which has changed
the contents of any files.
When restarting the node with a new version of
ndbd.
As a measure of last resort when for some reason the
node restart or system restart repeatedly fails. In
this case, be aware that this node can no longer be
used to restore data due to the destruction of the
data files.
Use of this option prevents the
StartPartialTimeout and
StartPartitionedTimeout configuration
parameters from having any effect.
Important
This option does not affect either
of the following:
This option also has no effect on recovery of data by a
data node that is just starting (or restarting) from
data nodes that are already running. This recovery of
data occurs automatically, and requires no user
intervention in a MySQL Cluster that is running
normally.
It is permissible to use this option when starting the
cluster for the very first time (that is, before any data
node files have been created); however, it is
not necessary to do so.
--initial-start
This option is used when performing a partial initial
start of the cluster. Each node should be started with
this option, as well as
--nowait-nodes .
Suppose that you have a 4-node cluster whose data nodes
have the IDs 2, 3, 4, and 5, and you wish to perform a
partial initial start using only nodes 2, 4, and 5 —
that is, omitting node 3:
shell> ndbd --ndb-nodeid=2 --nowait-nodes=3 --initial-start
shell> ndbd --ndb-nodeid=4 --nowait-nodes=3 --initial-start
shell> ndbd --ndb-nodeid=5 --nowait-nodes=3 --initial-start
Prior to MySQL 5.1.19, it was not possible to perform DDL
operations involving Disk Data tables on a partially
started cluster. (See Bug#24631.)
When using this option, you must also specify the node ID
for the data node being started with the
--ndb-nodeid option.
This option was added in MySQL 5.1.11.
Important
Do not confuse this option with the
--nowait-nodes option
added for ndb_mgmd in MySQL Cluster
NDB 7.0.10, which can be used to allow a cluster
configured with multiple management servers to be
started without all management servers being online.
--nowait-nodes=node_id_1 [,
node_id_2 [, ...]]
This option takes a list of data nodes which for which the
cluster will not wait for before starting.
This can be used to start the cluster in a partitioned
state. For example, to start the cluster with only half of
the data nodes (nodes 2, 3, 4, and 5) running in a 4-node
cluster, you can start each ndbd
process with --nowait-nodes=3,5 . In this
case, the cluster starts as soon as nodes 2 and 4 connect,
and does not wait
StartPartitionedTimeout milliseconds
for nodes 3 and 5 to connect as it would otherwise.
If you wanted to start up the same cluster as in the
previous example without one ndbd
— say, for example, that the host machine for node 3
has suffered a hardware failure — then start nodes
2, 4, and 5 with --nowait-nodes=3 . Then
the cluster will start as soon as nodes 2, 4, and 5
connect and will not wait for node 3 to start.
This option was added in MySQL 5.1.9.
--nodaemon
Instructs ndbd not to start as a daemon
process. This is useful when ndbd is
being debugged and you want output to be redirected to the
screen.
As of MySQL Cluster NDB 7.0.8, the default behavior for
ndbd and ndbmtd on
Windows is to run in the foreground, making this option
unnecessary on Windows platforms. (Bug#45588)
--nostart , -n
Instructs ndbd not to start
automatically. When this option is used,
ndbd connects to the management server,
obtains configuration data from it, and initializes
communication objects. However, it does not actually start
the execution engine until specifically requested to do so
by the management server. This can be accomplished by
issuing the proper START command in the
management client (see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”).
ndbd generates a set of log files which are
placed in the directory specified by
DataDir in the
config.ini configuration file.
These log files are listed below.
node_id is the node's unique
identifier. Note that node_id
represents the node's unique identifier. For example,
ndb_2_error.log is the error log
generated by the data node whose node ID is
2 .
ndb_node_id _error.log
is a file containing records of all crashes which the
referenced ndbd process has
encountered. Each record in this file contains a brief
error string and a reference to a trace file for this
crash. A typical entry in this file might appear as shown
here:
Date/Time: Saturday 30 July 2004 - 00:20:01
Type of error: error
Message: Internal program error (failed ndbrequire)
Fault ID: 2341
Problem data: DbtupFixAlloc.cpp
Object of reference: DBTUP (Line: 173)
ProgramName: NDB Kernel
ProcessID: 14909
TraceFile: ndb_2_trace.log.2
***EOM***
Listings of possible ndbd exit codes
and messages generated when a data node process shuts down
prematurely can be found in
ndbd Error Messages.
Important
The last entry in the error log file is not
necessarily the newest one (nor is it likely
to be). Entries in the error log are
not listed in chronological order;
rather, they correspond to the order of the trace files
as determined in the
ndb_node_id _trace.log.next
file (see below). Error log entries are thus overwritten
in a cyclical and not sequential fashion.
ndb_node_id _trace.log.trace_id
is a trace file describing exactly what happened just
before the error occurred. This information is useful for
analysis by the MySQL Cluster development team.
It is possible to configure the number of these trace
files that will be created before old files are
overwritten. trace_id is a
number which is incremented for each successive trace
file.
ndb_node_id _trace.log.next
is the file that keeps track of the next trace file number
to be assigned.
ndb_node_id _out.log
is a file containing any data output by the
ndbd process. This file is created only
if ndbd is started as a daemon, which
is the default behavior.
ndb_node_id .pid
is a file containing the process ID of the
ndbd process when started as a daemon.
It also functions as a lock file to avoid the starting of
nodes with the same identifier.
ndb_node_id _signal.log
is a file used only in debug versions of
ndbd, where it is possible to trace all
incoming, outgoing, and internal messages with their data
in the ndbd process.
It is recommended not to use a directory mounted through NFS
because in some environments this can cause problems whereby
the lock on the .pid file remains in
effect even after the process has terminated.
To start ndbd, it may also be necessary to
specify the host name of the management server and the port on
which it is listening. Optionally, one may also specify the
node ID that the process is to use.
shell> ndbd --connect-string="nodeid=2;host=ndb_mgmd.mysql.com:1186"
See Section 17.3.2.3, “The MySQL Cluster Connectstring”, for
additional information about this issue.
Section 17.4.2, “ndbd — The MySQL Cluster Data Node Daemon”, describes other
options for ndbd.
When ndbd starts, it actually initiates two
processes. The first of these is called the “angel
process”; its only job is to discover when the
execution process has been completed, and then to restart the
ndbd process if it is configured to do so.
Thus, if you attempt to kill ndbd via the
Unix kill command, it is necessary to kill
both processes, beginning with the angel process. The
preferred method of terminating an ndbd
process is to use the management client and stop the process
from there.
The execution process uses one thread for reading, writing,
and scanning data, as well as all other activities. This
thread is implemented asynchronously so that it can easily
handle thousands of concurrent actions. In addition, a
watch-dog thread supervises the execution thread to make sure
that it does not hang in an endless loop. A pool of threads
handles file I/O, with each thread able to handle one open
file. Threads can also be used for transporter connections by
the transporters in the ndbd process. In a
multi-processor system performing a large number of operations
(including updates), the ndbd process can
consume up to 2 CPUs if permitted to do so.
For a machine with many CPUs it is possible to use several
ndbd processes which belong to different
node groups; however, such a configuration is still considered
experimental and is not supported for MySQL 5.1
in a production setting. See
Section 17.1.5, “Known Limitations of MySQL Cluster”.
17.4.3. ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)
ndbmtd is a multi-threaded version of
ndbd, the process that is used to handle
all the data in tables using the
NDBCLUSTER storage engine.
ndbmtd is intended for use on host
computers having multiple CPU cores. Except where otherwise
noted, ndbmtd functions in the same way as
ndbd; therefore, in this section, we
concentrate on the ways in which ndbmtd
differs from ndbd, and you should consult
Section 17.4.2, “ndbd — The MySQL Cluster Data Node Daemon”, for additional
information about running MySQL Cluster data nodes that apply
to both the single-threaded and multi-threaded versions of the
data node process.
Command-line options and configuration parameters used with
ndbd also apply to
ndbmtd. For more information about these
options and parameters, see
Section 17.4.2, “ndbd — The MySQL Cluster Data Node Daemon”, and
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, respectively.
ndbmtd is also file system-compatible with
ndbd. In other words, a data node running
ndbd can be stopped, the binary replaced
with ndbmtd, and then restarted without any
loss of data. (However, when doing this, you must make sure
that MaxNoOfExecutionThreads is set to an
apppriate value before restarting the node if you wish for
ndbmtd to run in multi-threaded fashion.)
Similarly, an ndbmtd binary can be replaced
with ndbd simply by stopping the node and
then starting ndbd in place of the
multi-threaded binary. It is not necessary when switching
between the two to start the data node binary using
--initial .
Prior to MySQL Cluster NDB 7.0.6, there were known issues when
using ndbmtd with MySQL Cluster Disk Data
tables. If you wish to use multi-threaded data nodes with
disk-based NDB tables, you should ensure
that you are running MySQL Cluster NDB 7.0.6 or later.
(Bug#41915, Bug#44915)
Using ndbmtd differs from using
ndbd in two key respects:
You must set an appropriate value for the
MaxNoOfExecutionThreads configuration
parameter in the config.ini file. If
you do not do so, ndbmtd runs in
single-threaded mode — that is, it behaves like
ndbd.
Trace files are generated by critical errors in
ndbmtd processes in a somewhat
different fashion from how these are generated by
ndbd failures.
These differences are discussed in more detail in the next few
paragraphs.
Number of execution threads.
The MaxNoOfExecutionThreads configuration
parameter is used to determine the number of local query
handler (LQH) threads spawned by ndbmtd.
Although this parameter is set in [ndbd]
or [ndbd default] sections of the
config.ini file, it is exclusive to
ndbmtd and does not apply to
ndbd.
This parameter takes an integer value from 2 to 8 inclusive.
Generally, you should set this parameter equal to the number
of CPU cores on the data node host, as shown in the following
table:
(It is possible to set this parameter to other values within
the permitted range, but these are automatically rounded as
shown in the Value Used
column of the next table in this section.)
The multi-threaded data node process always spawns at least 4
threads:
1 local query handler (LQH) thread
1 transaction coordinator (TC) thread
1 transporter thread
1 subscription manager (SUMA) thread
Setting this parameter to a value between 4 and 8 inclusive
causes additional LQH threads to be used by
ndbmtd (up to a maximum of 4 LQH threads),
as shown in the following table:
Setting this parameter outside the permitted range of values
causes the management server to abort on startup with the
error Error line number :
Illegal value value for parameter
MaxNoOfExecutionThreads.
Note
In MySQL Cluster NDB 6.4.0, it is not possible to set
MaxNoOfExecutionThreads to 2. You can
safely use the value 3 instead (it is treated as 2
internally). This issue is resolved in MySQL Cluster NDB
6.4.1.
In MySQL Cluster NDB 6.4.0 through 6.4.3, the default value
for this parameter was undefined, although the default
behavior for ndbmtd was to use 1 LQH
thread, as though MaxNoOfExecutionThreads
had been set to 2. Beginning with MySQL Cluster NDB 7.0.4,
this parameter has an explcit default value of 2, thus
guaranteeing this default behavior.
In MySQL Cluster NDB 7.0, it is not possible to cause
ndbmtd to use more than 1 TC thread,
although we plan to introduce this capability in a future
MySQL Cluster release series.
Like ndbd, ndbmtd
generates a set of log files which are placed in the directory
specified by DataDir in the
config.ini configuration file. Except for
trace files, these are generated in the same way and have the
same names as those generated by ndbd.
In the event of a critical error, ndbmtd
generates trace files describing what happened just prior to
the error' occurrence. These files, which can be found in
the data node's DataDir , are useful
for analysis of problems by the MySQL Cluster Development and
Support teams. One trace file is generated for each
ndbmtd thread. The names of these files
follow the pattern
ndb_node_id _trace.log.trace_id _tthread_id ,
where node_id is the data
node's unique node ID in the cluster,
trace_id is a trace sequence
number, and thread_id is the thread
ID. For example, in the event of the failure of an
ndbmtd process running as a MySQL Cluster
data node having the node ID 3 and with
MaxNoOfExecutionThreads equal to 4, four
trace files are generated in the data node's data
directory; if the is the first time this node has failed, then
these files are named
ndb_3_trace.log.1_t1 ,
ndb_3_trace.log.1_t2 ,
ndb_3_trace.log.1_t3 , and
ndb_3_trace.log.1_t4 . Internally, these
trace files follow the same format as ndbd
trace files.
The ndbd exit codes and messages that are
generated when a data node process shuts down prematurely are
also used by ndbmtd. See
ndbd Error Messages, for a listing of these.
Note
It is possible to use ndbd and
ndbmtd concurrently on different data
nodes in the same MySQL Cluster. However, such
configurations have not been tested extensively; thus, we
cannot not recommend doing so in a production setting at
this time.
17.4.4. ndb_mgmd — The MySQL Cluster Management Server Daemon
The management server is the process that reads the cluster
configuration file and distributes this information to all
nodes in the cluster that request it. It also maintains a log
of cluster activities. Management clients can connect to the
management server and check the cluster's status.
The following table includes options that are specific to the
MySQL Cluster management server program
ndb_mgmd. Additional descriptions follow
the table. For options common to all MySQL Cluster programs,
see Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.10. ndb_mgmd Command Line
Options Format | Description | Introduction | Deprecated | Removed |
---|
--bind-address | Local bind address | 5.1.22-ndb-6.3.2 | | | -c | Specify the cluster configuration file; in NDB-6.4.0 and later, needs --reload or --initial to override configuration cache if present | | | | --configdir=directory | Specify the cluster management server's configuration cache directory | 5.1.30-ndb-6.4.0 | | | --daemon | Run ndb_mgmd in daemon mode (default) | | | | --initial | Causes the management server reload its configuration data from the configuration file, bypassing the configuration cache | 5.1.30-ndb-6.4.0 | | | --interactive | Run ndb_mgmd in interactive mode (not officially supported in production; for testing purposes only) | | | | --log-name= | A name to use when writing messages applying to this node in the cluster log. | 5.1.37-ndb-7.0.8 | | | --mycnf | Read cluster configuration data from the my.cnf file | | | | --no-nodeid-checks | Do not provide any node id checks | | | | --nodaemon | Do not run ndb_mgmd as a daemon | | | | --nowait-nodes=list | Do not wait for these management nodes when starting this management server. Also requires --ndb-nodeid to be used. | 5.1.39-ndb-7.1.0 | | | --print-full-config | Print full configuration and exit | | | | --reload | Causes the management server to compare the configuration file with its configuration cache | 5.1.30-ndb-6.4.0 | | |
--bind-address=host [:port ]
When specified, this option limits management server
connections by management clients to clients at the
specified host name or IP address (and possibly port, if
this is also specified). In such cases, a management
client attempting to connect to the management server from
any other address fails with the error Unable
to setup port:
host :port !
If the port is not specified,
the management client attempts to use port 1186.
This option was added in MySQL Cluster NDB 6.2.5 and MySQL
Cluster NDB 6.3.2.
--configdir=directory
Beginning with MySQL Cluster NDB 6.4.0, configuration data
is cached internally rather than being read from the
cluster global configuration file each time the management
server is started (see
Section 17.3.2, “MySQL Cluster Configuration Files”). This option
instructs the management server to its configuration cache
in the directory indicated. By
default, this is a directory named
mysql-cluster in the MySQL
installation directory — for example, if you compile
and install MySQL Cluster on a Unix system using the
default location, this is
/usr/local/mysql-cluster .
This behavior can be overridden using the
--initial or --reload
option for ndb_mgmd. Each of these
options is described elsewhere in this section.
This option is available beginning with MySQL Cluster NDB
6.4.0.
Beginning with MySQL Cluster NDB 7.0.8,
--config-dir is accepted as an alias for
--configdir .
--config-file=filename ,
-f filename
Instructs the management server as to which file it should
use for its configuration file. By default, the management
server looks for a file named
config.ini in the same directory as
the ndb_mgmd executable; otherwise the
file name and location must be specified explicitly.
Beginning with MySQL Cluster NDB 6.4.0, this option is
ignored unless the management server is forced to read the
configuration file, either because
ndb_mgmd was started with the
--reload or --initial
option, or because the management server could not find
any configuration cache. See
Section 17.3.2, “MySQL Cluster Configuration Files”, for more
information.
--daemon , -d
Instructs ndb_mgmd to start as a daemon
process. This is the default behavior.
This option has no effect when running
ndb_mgmd on Windows platforms.
--initial
Beginning with MySQL Cluster NDB 6.4.0, configuration data
is cached internally rather than being read from the
cluster global configuration file each time the management
server is started (see
Section 17.3.2, “MySQL Cluster Configuration Files”). Using this
option overrides this behavior, by forcing the management
server to delete any existing cache files, and then to
re-read the configuration data from the cluster
configuration file and to build a new cache.
This differs in two ways from the
--reload option. First,
--reload forces the server to check the
configuration file against the cache and reload its data
only if the contents of the file are different from the
cache. Second, --reload does not delete
any existing cache files.
If ndb_mgmd is invoked with
--initial but cannot find a global
configuration file, the management server cannot start.
This option was introduced in MySQL Cluster NDB 6.4.0.
--log-name=name
Provides a name to be used for this node in the cluster
log.
This option was added in MySQL Cluster NDB 7.0.8.
--nodaemon
Instructs ndb_mgmd not to start as a
daemon process.
As of MySQL Cluster NDB 7.0.8, the default behavior for
ndb_mgmd on Windows is to run in the
foreground, making this option unnecessary on Windows
platforms. (Bug#45588)
--print-full-config , -P
Shows extended information regarding the configuration of
the cluster. With this option on the command line the
ndb_mgmd process prints information
about the cluster setup including an extensive list of the
cluster configuration sections as well as parameters and
their values. Normally used together with the
--config-file (-f )
option.
--reload
Beginning with MySQL Cluster NDB 6.4.0, configuration data
is stored internally rather than being read from the
cluster global configuration file each time the management
server is started (see
Section 17.3.2, “MySQL Cluster Configuration Files”). Using this
option forces the management server to check its internal
data store against the cluster configuration file and to
reload the configuration if it finds that the
configuration file does not match the cache. Existing
configuration cache files are preserved, but not used.
This differs in two ways from the
--initial option. First,
--initial causes all cache files to be
deleted. Second, --initial forces the
management server to re-read the global configuration file
and construct a new cache.
If the management server cannot find a global
configuration file, then the --reload
option is ignored.
This option was introduced in MySQL Cluster NDB 6.4.0.
--nowait-nodes
When starting a MySQL Cluster is configured with two
management nodes and running MySQL Cluster NDB 7.0 and
later, each management server normally checks to see
whether the other ndb_mgmd is also
operational and whether the other management server's
configuration is identical to its own. However, it is
sometimes desirable to start the cluster with only one
management node (and perhaps to allow the other
ndb_mgmd to be started later). This
option causes the management node to bypass any checks for
any other management nodes whose node IDs are passed to
this option, allowing the cluster to start as though
configured to use only the management node that was
started.
For purposes of illustration, consider the following
portion of a config.ini file (where
we have omitted most of the configuration parameters that
are not relevant to this example):
[ndbd]
NodeId = 1
HostName = 192.168.0.101
[ndbd]
NodeId = 2
HostName = 192.168.0.102
[ndbd]
NodeId = 3
HostName = 192.168.0.103
[ndbd]
NodeId = 4
HostName = 192.168.0.104
[mgm]
NodeId = 10
HostName = 192.168.0.150
[mgm]
NodeId = 11
HostName = 192.168.0.151
[api]
NodeId = 20
HostName = 192.168.0.200
[api]
NodeId = 21
HostName = 192.168.0.201
Assume that you wish to start this cluster using only the
management server having node ID 10 and
running on the host having the IP address 192.168.0.150.
(Suppose, for example, that the host computer on which you
intend to the other management server is temporarily
unavailable due to a hardware failure, and you are waiting
for it to be repaired.) To start the cluster in this way,
use a command line on the machine at 192.168.0.150 to
enter the following command:
shell> ndb_mgmd --ndb-nodeid=10 --nowait-nodes=11
As shown in the preceding example, when using
--nowait-nodes , you must
also use the
--ndb-nodeid option to
specify the node ID of this ndb_mgmd
process.
You can then start each of the cluster's data nodes
in the usual way. If you wish to start and use the second
management server in addition to the first management
server at a later time without restarting the data nodes,
you must start each data node with a connectstring that
references both management servers, like this:
shell> ndbd -c 192.168.0.150,192.168.0.151
The same is true with regard to the connectstring used
with any mysqld processes that you wish
to start as MySQL Cluster SQL nodes connected to this
cluster. See
Section 17.3.2.3, “The MySQL Cluster Connectstring”, for more
information.
When used with ndb_mgmd, this option
affects the behavior of the management node with regard to
other management nodes only. Do not confuse it with the
--nowait-nodes option used
with ndbd (or ndbmtd
in MySQL Cluster NDB 7.0 and later) to allow a cluster to
start with fewer than its full complement of data nodes;
when used with data nodes, this option affects their
behavior only with regard to other data nodes.
Multiple management node IDs may be passed to this option
as a comma-separated list. Each node ID must be no less
than 1 and no greater than 255. In practice, it is quite
rare to use more than two management servers for the same
MySQL Cluster (or to have any need for doing so); in most
cases you need to pass to this option only the single node
ID for the one management server that you do not wish to
use when starting the cluster.
Note
When you later start the “missing”
management server, its configuration must match that of
the management server that is already in use by the
cluster. Otherwise, it fails the configuration check
performed by the existing management server, and does
not start.
This option was introduced in MySQL Cluster NDB 7.0.10.
It is not strictly necessary to specify a connectstring when
starting the management server. However, if you are using more
than one management server, a connectstring should be provided
and each node in the cluster should specify its node ID
explicitly.
See Section 17.3.2.3, “The MySQL Cluster Connectstring”, for
information about using connectstrings.
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”, describes
other options for ndb_mgmd.
The following files are created or used by
ndb_mgmd in its starting directory, and are
placed in the DataDir as specified in the
config.ini configuration file. In the
list that follows, node_id is the
unique node identifier.
config.ini is the configuration file
for the cluster as a whole. This file is created by the
user and read by the management server.
Section 17.3, “MySQL Cluster Configuration”, discusses
how to set up this file.
ndb_node_id _cluster.log
is the cluster events log file. Examples of such events
include checkpoint startup and completion, node startup
events, node failures, and levels of memory usage. A
complete listing of cluster events with descriptions may
be found in Section 17.5, “Management of MySQL Cluster”.
When the size of the cluster log reaches one million
bytes, the file is renamed to
ndb_node_id _cluster.log.seq_id ,
where seq_id is the sequence
number of the cluster log file. (For example: If files
with the sequence numbers 1, 2, and 3 already exist, the
next log file is named using the number
4 .)
ndb_node_id _out.log
is the file used for stdout and
stderr when running the management
server as a daemon.
ndb_node_id .pid
is the process ID file used when running the management
server as a daemon.
17.4.5. ndb_mgm — The MySQL Cluster Management Client
The ndb_mgm management client process is
actually not needed to run the cluster. Its value lies in
providing a set of commands for checking the cluster's status,
starting backups, and performing other administrative
functions. The management client accesses the management
server using a C API. Advanced users can also employ this API
for programming dedicated management processes to perform
tasks similar to those performed by
ndb_mgm.
To start the management client, it is necessary to supply the
host name and port number of the management server:
shell> ndb_mgm [host_name [port_num ]]
For example:
shell> ndb_mgm ndb_mgmd.mysql.com 1186
The default host name and port number are
localhost and 1186, respectively.
The following table includes options that are specific to the
MySQL Cluster management client program
ndb_mgm. Additional descriptions follow the
table. For options common to all MySQL Cluster programs, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.11. ndb_mgm Command Line
Options Format | Description | Introduction | Deprecated | Removed |
---|
--try-reconnect=# | Specify number of tries for connecting to ndb_mgmd (0 = infinite) | | | |
--try-reconnect=number
If the connection to the management server is broken, the
node tries to reconnect to it every 5 seconds until it
succeeds. By using this option, it is possible to limit
the number of attempts to
number before giving up and
reporting an error instead.
Additional information about using ndb_mgm
can be found in
Section 17.5.2, “Commands in the MySQL Cluster Management Client”.
17.4.6. ndb_config — Extract MySQL Cluster Configuration Information
This tool extracts current configuration information for data
nodes, SQL nodes, and API nodes from a cluster management node
(and possibly its config.ini file).
Beginning with MySQL Cluster NDB 6.3.25 and MySQL Cluster NDB
7.0.6, it can also provide an offline dump (in text or XML
format) of all configuration parameters which can be used,
along with their default, maximum, and minimum values and
other information (see the discussion of the
--configinfo and --xml
options later in this section).
The following table includes options that are specific to
ndb_config. Additional descriptions follow
the table. For options common to all MySQL Cluster programs,
see Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.12. ndb_config Command Line
Options Format | Description | Introduction | Deprecated | Removed |
---|
--configinfo | Dumps information about all NDB configuration parameters in text format with default, maximum, and minimum values. Use with --xml to obtain XML output. | 5.1.34-ndb-7.0.6 | | | --connections | Print connection information only | | | | --fields=string | Field separator | | | | --host=name | Specify host | | | | --mycnf | Read configuration data from my.cnf file | | | | --nodeid | Get configuration of node with this ID | | | | --nodes | Print node information only | | | | | Short form for --ndb-connectstring | 5.1.12 | | | --config-file=path | Set the path to config.ini file | | | | --query=string | One or more query options (attributes) | | | | --rows=string | Row separator | | | | --type=name | Specify node type | | | | --configinfo --xml | Use with --configinfo to obtain a dump of all NDB configuration parameters in XML format with default, maximum, and minimum values. | 5.1.34-ndb-7.0.6 | | |
--usage , --help , or
-?
Causes ndb_config to print a list of
available options, and then exit.
--version , -V
Causes ndb_config to print a version
information string, and then exit.
--ndb-connectstring=connect_string
Specifies the connectstring to use in connecting to the
management server. The format for the connectstring is the
same as described in
Section 17.3.2.3, “The MySQL Cluster Connectstring”, and
defaults to localhost:1186 .
The use of -c as a short version for this
option is supported for ndb_config
beginning with MySQL 5.1.12.
--config-file=path-to-file
Gives the path to the management server's configuration
file (config.ini ). This may be a
relative or absolute path. If the management node resides
on a different host from the one on which
ndb_config is invoked, then an absolute
path must be used.
--query=query-options ,
-q
query-options
This is a comma-delimited list of query
options — that is, a list of one or more
node attributes to be returned. These include
id (node ID), type (node type —
that is, ndbd ,
mysqld , or
ndb_mgmd ), and any configuration
parameters whose values are to be obtained.
For example,
--query=id,type,indexmemory,datamemory
would return the node ID, node type,
DataMemory , and
IndexMemory for each node.
Note
If a given parameter is not applicable to a certain type
of node, than an empty string is returned for the
corresponding value. See the examples later in this
section for more information.
--host=hostname
Specifies the host name of the node for which
configuration information is to be obtained.
--id=node_id ,
--nodeid=node_id
Used to specify the node ID of the node for which
configuration information is to be obtained.
--nodes
(Tells ndb_config to print information
from parameters defined in [ndbd]
sections only. Currently, using this option has no affect,
since these are the only values checked, but it may become
possible in future to query parameters set in
[tcp] and other sections of cluster
configuration files.)
--type=node_type
Filters results so that only configuration values applying
to nodes of the specified
node_type
(ndbd , mysqld , or
ndb_mgmd ) are returned.
--fields=delimiter ,
-f delimiter
Specifies a delimiter string
used to separate the fields in the result. The default is
“, ” (the comma character).
Note
If the delimiter contains
spaces or escapes (such as \n for the
linefeed character), then it must be quoted.
--rows=separator ,
-r separator
Specifies a separator string
used to separate the rows in the result. The default is a
space character.
Note
If the separator contains
spaces or escapes (such as \n for the
linefeed character), then it must be quoted.
--configinfo
The --configinfo option, added in MySQL
Cluster NDB 6.3.25 and MySQL Cluster NDB 7.0.6, causes
ndb_config to dump a list of each MySQL
Cluster configuration parameter supported by the MySQL
Cluster distribution of which
ndb_config is a part, including the
following information:
A brief description of each parameter's purpose,
effects, and usage
The section of the config.ini
file where the parameter may be used
The parameter's data type or unit of measurement
Where applicable, the parameter's default,
minimum, and maximum values
A brief description of the parameter's purpose,
effects, and usage
MySQL Cluster release version and build information
By default, this output is in text format. Part of this
output is shown here:
shell> ndb_config --configinfo
****** SYSTEM ******
Name (String)
Name of system (NDB Cluster)
MANDATORY
PrimaryMGMNode (Non-negative Integer)
Node id of Primary ndb_mgmd(MGM) node
Default: 0 (Min: 0, Max: 4294967039)
ConfigGenerationNumber (Non-negative Integer)
Configuration generation number
Default: 0 (Min: 0, Max: 4294967039)
****** DB ******
MaxNoOfSubscriptions (Non-negative Integer)
Max no of subscriptions (default 0 == MaxNoOfTables)
Default: 0 (Min: 0, Max: 4294967039)
MaxNoOfSubscribers (Non-negative Integer)
Max no of subscribers (default 0 == 2 * MaxNoOfTables)
Default: 0 (Min: 0, Max: 4294967039)
…
--configinfo --xml
You can obtain the output of ndb_config
--configinfo as XML by adding the
--xml option (like the
--configinfo option, available beginning
with MySQL Cluster NDB 6.3.25 and MySQL Cluster NDB
7.0.6). A portion of the resulting output is shown in this
example:
shell> ndb_config --configinfo --xml
<configvariables protocolversion="1" ndbversionstring="mysql-5.1.34 ndb-7.0.6"
ndbversion="458758" ndbversionmajor="7" ndbversionminor="0"
ndbversionbuild="6">
<section name="SYSTEM">
<param name="Name" comment="Name of system (NDB Cluster)" type="string"
mandatory="true"/>
<param name="PrimaryMGMNode" comment="Node id of Primary ndb_mgmd(MGM) node"
type="unsigned" default="0" min="0" max="4294967039"/>
<param name="ConfigGenerationNumber" comment="Configuration generation number"
type="unsigned" default="0" min="0" max="4294967039"/>
</section>
<section name="NDBD">
<param name="MaxNoOfSubscriptions" comment="Max no of subscriptions (default 0 == MaxNoOfTables)"
type="unsigned" default="0" min="0" max="4294967039"/>
<param name="MaxNoOfSubscribers" comment="Max no of subscribers (default 0 == 2 * MaxNoOfTables)"
type="unsigned" default="0" min="0" max="4294967039"/>
…
</section>
…
</configvariables>
Note
Normally, the XML output produced by
ndb_config
--configinfo --xml is
formatted using one line per element; we have added
extra whitespace in the previous example, as well as the
next one, for reasons of legibility. This should not
make any difference to applications using this output,
since most XML processors either ignore nonessential
whitespace as a matter of course, or can be instructed
to do so.
Beginning with MySQL Cluster NDB 6.3.29 and MySQL Cluster
NDB 7.0.10, the XML output also indicates when changing a
given parameter requires that nodes be restarted using the
--initial option. This is shown by the
presence of an initial="true" attribute
in the corresponding <param>
element. In addition (also beginning with MySQL Cluster
NDB 6.3.29 and MySQL Cluster NDB 7.0.10), the restart type
(system or node ) is
also shown; if a given parameter requires a system
restart, this is indicated by the presence of a
restart="system" attribute in the
corresponding <param> element.
For example, the Diskless parameter
requires a system initial restart, as shown here (with the
restart and initial
attributes highlighted for visibility):
<param name="Diskless" comment="Run wo/ disk" type="bool" default="false"
restart="system" initial="true"/>
Currently, no initial attribute is
included in the XML output for
<param> elements corresponding to
parameters which do not require initial restarts; in other
words, initial="false" is the default,
and the value false should be assumed
if the attribute is not present. Similarly, the default
restart type is node (that is, an
online or “rolling” restart of the cluster),
but the restart attribute is included
only if the restart type is system
(meaning that all cluster nodes must be shut down at the
same time, then restarted).
Important
The --xml option can be used only with
the --configinfo option. Using
--xml without
--configinfo fails with an error.
Unlike the options used with this program to obtain
current configuration data, --configinfo
and --xml use information obtained from
the MySQL Cluster sources when
ndb_config was compiled. For this
reason, no connection to a running MySQL Cluster or access
to a config.ini or
my.cnf file is required for these two
options.
Combining other ndb_config options
(such as --query or
--type ) with
--configinfo or --xml is
not supported. If you attempt to do so, the usual
(current) result is that all other options besides
--configinfo or --xml
are simply ignored. However, this behavior is not
guaranteed and is subject to change at any time. In
addition, since ndb_config when used
with the --configinfo option does not
access the MySQL Cluster or read any files, trying to
specify additional options such as
--ndb-connectstring or
--config-file with
--configinfo serves no purpose.
Examples:
To obtain the node ID and type of each node in the
cluster:
shell> ./ndb_config --query=id,type --fields=':' --rows='\n'
1:ndbd
2:ndbd
3:ndbd
4:ndbd
5:ndb_mgmd
6:mysqld
7:mysqld
8:mysqld
9:mysqld
In this example, we used the --fields
options to separate the ID and type of each node with a
colon character (: ), and the
--rows options to place the values for
each node on a new line in the output.
To produce a connectstring that can be used by data, SQL,
and API nodes to connect to the management server:
shell> ./ndb_config --config-file=usr/local/mysql/cluster-data/config.ini --query=hostname,portnumber --fields=: --rows=, --type=ndb_mgmd
192.168.0.179:1186
This invocation of ndb_config checks
only data nodes (using the --type
option), and shows the values for each node's ID and host
name, and its DataMemory ,
IndexMemory , and
DataDir parameters:
shell> ./ndb_config --type=ndbd --query=id,host,datamemory,indexmemory,datadir -f ' : ' -r '\n'
1 : 192.168.0.193 : 83886080 : 18874368 : /usr/local/mysql/cluster-data
2 : 192.168.0.112 : 83886080 : 18874368 : /usr/local/mysql/cluster-data
3 : 192.168.0.176 : 83886080 : 18874368 : /usr/local/mysql/cluster-data
4 : 192.168.0.119 : 83886080 : 18874368 : /usr/local/mysql/cluster-data
In this example, we used the short options
-f and -r for setting
the field delimiter and row separator, respectively.
To exclude results from any host except one in particular,
use the --host option:
shell> ./ndb_config --host=192.168.0.176 -f : -r '\n' -q id,type
3:ndbd
5:ndb_mgmd
In this example, we also used the short form
-q to determine the attributes to be
queried.
Similarly, you can limit results to a node with a specific
ID using the --id or
--nodeid option.
17.4.7. ndb_cpcd — Automate Testing for NDB Development
This utility is found in the libexec
directory. It is part of an internal automated test framework
used in testing and debugging MySQL Cluster. Because it can
control processes on remote systems, it is not advisable to
use ndb_cpcd in a production cluster.
The source files for ndb_cpcd may be found
in the directory storage/ndb/src/cw/cpcd ,
in the MySQL Cluster source tree.
17.4.8. ndb_delete_all — Delete All Rows from an NDB Table
ndb_delete_all deletes all rows from the
given NDB table. In some cases,
this can be much faster than
DELETE or even
TRUNCATE TABLE .
Usage:
ndb_delete_all -c connect_string tbl_name -d db_name
This deletes all rows from the table named
tbl_name in the database named
db_name . It is exactly equivalent
to executing TRUNCATE
db_name .tbl_name
in MySQL.
Additional Options:
17.4.9. ndb_desc — Describe NDB Tables
ndb_desc provides a detailed description of
one or more NDB tables.
Usage:
ndb_desc -c connect_string tbl_name -d db_name [-p]
Sample Output:
MySQL table creation and population statements:
USE test;
CREATE TABLE fish (
id INT(11) NOT NULL AUTO_INCREMENT,
name VARCHAR(20) NOT NULL,
length_mm INT(11) NOT NULL,
weight_gm INT(11) NOT NULL,
PRIMARY KEY pk (id),
UNIQUE KEY uk (name)
) ENGINE=NDB;
INSERT INTO fish VALUES
('','guppy', 35, 2), ('','tuna', 2500, 150000),
('','shark', 3000, 110000), ('','manta ray', 1500, 50000),
('','grouper', 900, 125000), ('','puffer', 250, 2500);
Output from ndb_desc:
shell> ./ndb_desc -c localhost fish -d test -p
-- fish --
Version: 2
Fragment type: 9
K Value: 6
Min load factor: 78
Max load factor: 80
Temporary table: no
Number of attributes: 4
Number of primary keys: 1
Length of frm data: 311
Row Checksum: 1
Row GCI: 1
SingleUserMode: 0
ForceVarPart: 1
FragmentCount: 2
TableStatus: Retrieved
-- Attributes --
id Int PRIMARY KEY DISTRIBUTION KEY AT=FIXED ST=MEMORY AUTO_INCR
name Varchar(20;latin1_swedish_ci) NOT NULL AT=SHORT_VAR ST=MEMORY
length_mm Int NOT NULL AT=FIXED ST=MEMORY
weight_gm Int NOT NULL AT=FIXED ST=MEMORY
-- Indexes --
PRIMARY KEY(id) - UniqueHashIndex
PRIMARY(id) - OrderedIndex
uk$unique(name) - UniqueHashIndex
uk(name) - OrderedIndex
-- Per partition info --
Partition Row count Commit count Frag fixed memory Frag varsized memory Extent_space Free extent_space
0 2 2 32768 32768 0 0
1 4 4 32768 32768 0 0
NDBT_ProgramExit: 0 - OK
The Extent_space and Free
extent_space columns were added in MySQL Cluster NDB
6.3.27 and MySQL Cluster NDB 7.0.8. They are applicable only
to NDB tables having columns on disk; for
tables having only in-memory columns, these columns always
contain the value 0 .
To illustrate their use, we modify the previous example.
First, we must create the necessary Disk Data objects, as
shown here:
CREATE LOGFILE GROUP lg_1
ADD UNDOFILE 'undo_1.log'
INITIAL_SIZE 16M
UNDO_BUFFER_SIZE 2M
ENGINE NDB;
ALTER LOGFILE GROUP lg_1
ADD UNDOFILE 'undo_2.log'
INITIAL_SIZE 12M
ENGINE NDB;
CREATE TABLESPACE ts_1
ADD DATAFILE 'data_1.dat'
USE LOGFILE GROUP lg_1
INITIAL_SIZE 32M
ENGINE NDB;
ALTER TABLESPACE ts_1
ADD DATAFILE 'data_2.dat'
INITIAL_SIZE 48M
ENGINE NDB;
(For more information on the statements just shown and the
objects created by them, see
Section 17.5.10.1, “MySQL Cluster Disk Data Objects”, as well as
Section 12.1.14, “CREATE LOGFILE GROUP Syntax”, and
Section 12.1.18, “CREATE TABLESPACE Syntax”.)
Now we can create and populate a version of the
fish table that stores 2 of its columns on
disk (deleting the previous version of the table first, if it
already exists):
CREATE TABLE fish (
id INT(11) NOT NULL AUTO_INCREMENT,
name VARCHAR(20) NOT NULL,
length_mm INT(11) NOT NULL,
weight_gm INT(11) NOT NULL,
PRIMARY KEY pk (id),
UNIQUE KEY uk (name)
) TABLESPACE ts_1 STORAGE DISK
ENGINE=NDB;
INSERT INTO fish VALUES
('','guppy', 35, 2), ('','tuna', 2500, 150000),
('','shark', 3000, 110000), ('','manta ray', 1500, 50000),
('','grouper', 900, 125000), ('','puffer', 250, 2500);
When run against this version of the table,
ndb_desc displays the following output:
shell> ./ndb_desc -c localhost fish -d test -p
-- fish --
Version: 3
Fragment type: 9
K Value: 6
Min load factor: 78
Max load factor: 80
Temporary table: no
Number of attributes: 4
Number of primary keys: 1
Length of frm data: 321
Row Checksum: 1
Row GCI: 1
SingleUserMode: 0
ForceVarPart: 1
FragmentCount: 2
TableStatus: Retrieved
-- Attributes --
id Int PRIMARY KEY DISTRIBUTION KEY AT=FIXED ST=MEMORY AUTO_INCR
name Varchar(20;latin1_swedish_ci) NOT NULL AT=SHORT_VAR ST=MEMORY
length_mm Int NOT NULL AT=FIXED ST=DISK
weight_gm Int NOT NULL AT=FIXED ST=DISK
-- Indexes --
PRIMARY KEY(id) - UniqueHashIndex
PRIMARY(id) - OrderedIndex
uk$unique(name) - UniqueHashIndex
uk(name) - OrderedIndex
-- Per partition info --
Partition Row count Commit count Frag fixed memory Frag varsized memory Extent_space Free extent_space
0 2 2 32768 32768 1048576 1044440
1 4 4 32768 32768 1048576 1044400
NDBT_ProgramExit: 0 - OK
This means that 1048576 bytes are allocated from the
tablespace for this table on each partition, of which 1044440
bytes remain free for additional storage. In other words,
1048576 - 1044440 = 4136 bytes per partition is currently
being used to store the data from this table's disk-based
columns. The number of bytes shown as Free
extent_space is available for storing on-disk column
data from the fish table only; for this
reason, it is not visible when selecting from the
INFORMATION_SCHEMA.FILES table.
Additional Options:
--extra-partition-info ,
-p
Prints additional information about the table's
partitions.
Information about multiple tables can be obtained in a
single invocation of ndb_desc by using
their names, separated by spaces. All of the tables must
be in the same database.
17.4.10. ndb_drop_index — Drop Index from an NDB Table
ndb_drop_index drops the specified index
from an NDB table. It
is recommended that you use this utility only as an example
for writing NDB API applications — see the
Warning later in this section for details.
Usage:
ndb_drop_index -c connect_string table_name index -d db_name
The statement shown above drops the index named
index from the
table in the
database .
Additional Options: None that
are specific to this application.
Warning
Operations performed on Cluster table indexes
using the NDB API are not visible to MySQL and make the
table unusable by a MySQL server. If you use this
program to drop an index, then try to access the table from
an SQL node, an error results, as shown here:
shell> ./ndb_drop_index -c localhost dogs ix -d ctest1
Dropping index dogs/idx...OK
NDBT_ProgramExit: 0 - OK
shell> ./mysql -u jon -p ctest1
Enter password: *******
Reading table information for completion of table and column names
You can turn off this feature to get a quicker startup with -A
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 7 to server version: 5.1.41-ndb-7.0.12
Type 'help;' or '\h' for help. Type '\c' to clear the buffer.
mysql> SHOW TABLES;
+------------------+
| Tables_in_ctest1 |
+------------------+
| a |
| bt1 |
| bt2 |
| dogs |
| employees |
| fish |
+------------------+
6 rows in set (0.00 sec)
mysql> SELECT * FROM dogs;
ERROR 1296 (HY000): Got error 4243 'Index not found' from NDBCLUSTER
In such a case, your only option for
making the table available to MySQL again is to drop the table
and re-create it. You can use either the SQL
statementDROP TABLE or the
ndb_drop_table utility (see
Section 17.4.11, “ndb_drop_table — Drop an NDB Table”) to
drop the table.
17.4.11. ndb_drop_table — Drop an NDB Table
ndb_drop_table drops the specified
NDB table. (If you try to use
this on a table created with a storage engine other than NDB,
it fails with the error 723: No such table
exists.) This operation is extremely fast —
in some cases, it can be an order of magnitude faster than
using DROP TABLE on an
NDB table from MySQL.
Usage:
ndb_drop_table -c connect_string tbl_name -d db_name
Additional Options: None.
17.4.12. ndb_error_reporter — NDB Error-Reporting Utility
ndb_error_reporter creates an archive from
data node and management node log files that can be used to
help diagnose bugs or other problems with a cluster.
It is highly recommended that you make use of this
utility when filing reports of bugs in MySQL
Cluster.
Usage:
ndb_error_reporter path/to/config-file [username ] [--fs]
This utility is intended for use on a management node host,
and requires the path to the management host configuration
file (config.ini ). Optionally, you can
supply the name of a user that is able to access the cluster's
data nodes via SSH, in order to copy the data node log files.
ndb_error_reporter then includes all of these files in archive
that is created in the same directory in which it is run. The
archive is named
ndb_error_report_YYYYMMDDHHMMSS .tar.bz2 ,
where YYYYMMDDHHMMSS is a datetime
string.
If the --fs is used, then the data node file
systems are also copied to the management host and included in
the archive that is produced by this script. As data node file
systems can be extremely large even after being compressed, we
ask that you please do not send archives
created using this option to Sun Microsystems, Inc. unless you
are specifically requested to do so.
17.4.13. ndb_print_backup_file — Print NDB Backup File Contents
ndb_print_backup_file obtains diagnostic
information from a cluster backup file.
Usage:
ndb_print_backup_file file_name
file_name is the name of a cluster
backup file. This can be any of the files
(.Data , .ctl , or
.log file) found in a cluster backup
directory. These files are found in the data node's backup
directory under the subdirectory
BACKUP-# ,
where # is the sequence number for
the backup. For more information about cluster backup files
and their contents, see
Section 17.5.3.1, “MySQL Cluster Backup Concepts”.
Like ndb_print_schema_file and
ndb_print_sys_file (and unlike most of the
other NDB utilities that are
intended to be run on a management server host or to connect
to a management server)
ndb_print_backup_file must be run on a
cluster data node, since it accesses the data node file system
directly. Because it does not make use of the management
server, this utility can be used when the management server is
not running, and even when the cluster has been completely
shut down.
Additional Options: None.
17.4.14. ndb_print_schema_file — Print NDB Schema File Contents
ndb_print_schema_file obtains diagnostic
information from a cluster schema file.
Usage:
ndb_print_schema_file file_name
file_name is the name of a cluster
schema file. For more information about cluster schema files,
see Cluster Data Node FileSystemDir Files.
Like ndb_print_backup_file and
ndb_print_sys_file (and unlike most of the
other NDB utilities that are
intended to be run on a management server host or to connect
to a management server)
ndb_schema_backup_file must be run on a
cluster data node, since it accesses the data node file system
directly. Because it does not make use of the management
server, this utility can be used when the management server is
not running, and even when the cluster has been completely
shut down.
Additional Options: None.
17.4.15. ndb_print_sys_file — Print NDB System File Contents
ndb_print_sys_file obtains diagnostic
information from a MySQL Cluster system file.
Usage:
ndb_print_sys_file file_name
file_name is the name of a cluster
system file (sysfile). Cluster system files are located in a
data node's data directory (DataDir ); the
path under this directory to system files matches the pattern
ndb_# _fs/D# /DBDIH/P# .sysfile .
In each case, the # represents a
number (not necessarily the same number). For more
information, see
Cluster Data Node FileSystemDir Files.
Like ndb_print_backup_file and
ndb_print_schema_file (and unlike most of
the other NDB utilities that are
intended to be run on a management server host or to connect
to a management server)
ndb_print_backup_file must be run on a
cluster data node, since it accesses the data node file system
directly. Because it does not make use of the management
server, this utility can be used when the management server is
not running, and even when the cluster has been completely
shut down.
Additional Options: None.
17.4.16. ndbd_redo_log_reader — Check and Print Content of Cluster Redo Log
Reads a redo log file, checking it for errors, printing its
contents in a human-readable format, or both.
ndbd_redo_log_reader is intended for use
primarily by MySQL developers and support personnel in
debugging and diagnosing problems.
This utility was made available as part of default builds
beginning with MySQL Cluster NDB 6.1.3. It remains under
development, and its syntax and behavior are subject to change
in future releases. For this reason, it should be considered
experimental at this time.
The C++ source files for
ndbd_redo_log_reader can be found in the
directory
/storage/ndb/src/kernel/blocks/dblqh/redoLogReader .
The following table includes options that are specific to the
MySQL Cluster program ndbd_redo_log_reader.
Additional descriptions follow the table. For options common
to all MySQL Cluster programs, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.13. ndbd_redo_log_reader Command
Line Options Format | Description | Introduction | Deprecated | Removed |
---|
-nocheck | Do not check records for errors | | | | -noprint | Do not print records | | | |
Usage:
ndbd_redo_log_reader file_name [options ]
file_name is the name of a cluster
REDO log file. REDO log files are located in the numbered
directories under the data node's data directory
(DataDir ); the path under this directory to
the REDO log files matches the pattern
ndb_# _fs/D# /LCP/# /T# F# .Data .
In each case, the # represents a
number (not necessarily the same number). For more
information, see
Cluster Data Node FileSystemDir Files.
The name of the file to be read may be followed by one or more
of the options listed here:
Like ndb_print_backup_file and
ndb_print_schema_file (and unlike most of
the NDB utilities that are
intended to be run on a management server host or to connect
to a management server)
ndbd_redo_log_reader must be run on a
cluster data node, since it accesses the data node file system
directly. Because it does not make use of the management
server, this utility can be used when the management server is
not running, and even when the cluster has been completely
shut down.
17.4.17. ndb_restore — Restore a MySQL Cluster Backup
The cluster restoration program is implemented as a separate
command-line utility ndb_restore, which can
normally be found in the MySQL bin
directory. This program reads the files created as a result of
the backup and inserts the stored information into the
database.
ndb_restore must be executed once for each
of the backup files that were created by the START
BACKUP command used to create the backup (see
Section 17.5.3.2, “Using The MySQL Cluster Management Client to Create a Backup”).
This is equal to the number of data nodes in the cluster at
the time that the backup was created.
Note
Before using ndb_restore, it is
recommended that the cluster be running in single user mode,
unless you are restoring multiple data nodes in parallel.
See Section 17.5.6, “MySQL Cluster Single User Mode”, for
more information about single user mode.
The following table includes options that are specific to the
MySQL Cluster native backup restoration program
ndb_restore. Additional descriptions follow
the table. For options common to all MySQL Cluster programs,
see Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.14. ndb_restore Command Line
Options Format | Description | Introduction | Deprecated | Removed |
---|
--append | Append data to a tab-delimited file | 5.1.18 | | | --backup_path=path | Path to backup files directory | 5.1.17 | | | --backupid=# | Restore from the backup with the given ID | | | | --connect | Same as connectstring | | | | --restore_data | Restore table data and logs into NDB Cluster using the NDB API | | | | --dont_ignore_systab_0 | Do not ignore system table during restore. Experimental only; not for production use | | | | --exclude-databases=db-list | List of one or more databases to exclude (includes those not named) | 5.1.32-ndb-6.4.3 | | | --exclude-missing-columns | Causes columns from the backup version of a table that are missing from the version of the table in the database to be ignored. | 5.1.35-ndb-7.0.7 | | | --exclude-tables=table-list | List of one or more tables to exclude (includes those in same database that are not not named); each table reference must include the database name | 5.1.32-ndb-6.4.3 | | | --fields-enclosed-by=char | Fields are enclosed with the indicated character | 5.1.18 | | | --fields-optionally-enclosed-by | Fields are optionally enclosed with the indicated character | 5.1.18 | | | --fields-terminated-by=char | Fields are terminated by the indicated character | 5.1.18 | | | --hex | Print binary types in hexadecimal format | 5.1.18 | | | --include-databases=db-list | List of one or more databases to restore (excludes those not named) | 5.1.32-ndb-6.4.3 | | | --include-tables=table-list | List of one or more tables to restore (excludes those in same database that are not named); each table reference must include the database name | 5.1.32-ndb-6.4.3 | | | --lines-terminated-by=char | Lines are terminated by the indicated character | 5.1.18 | | | --restore_meta | Restore metadata to NDB Cluster using the NDB API | | | | --ndb-nodegroup-map=map | Nodegroup map for NDBCLUSTER storage engine. Syntax: list of (source_nodegroup, destination_nodegroup) | | | | --no-binlog | If a mysqld is connected and using binary logging, do not log the restored data | 5.1.24-ndb-6.3.16 | | | --no-restore-disk-objects | Do not restore Disk Data objects such as tablespaces and log file groups | | | | --no-upgrade | Do not upgrade array type for varsize attributes which do not already resize VAR data, and do not change column attributes | 5.1.19 | | | --nodeid=# | Back up files from node with this ID | | | | --parallelism=# | Number of parallel transactions during restoration of data | | | | --preserve-trailing-spaces | Allow preservation of tailing spaces (including padding) when CHAR is promoted to VARCHAR or BINARY is promoted to VARBINARY | 5.1.23-ndb-6.3.8 | | | --print | Print metadata, data and log to stdout (equivalent to --print_meta --print_data --print_log) | | | | --print_data | Print data to stdout | | | | --print_log | Print to stdout | | | | --print_metadata | Print metadata to stdout | | | | --progress-frequency=# | Print status of restoration each given number of seconds | | | | --promote-attributes | Allow attributes to be promoted when restoring data from backup | 5.1.23-ndb-6.3.8 | | | --restore_epoch | Restore epoch info into the status table. Convenient on a MySQL Cluster replication slave for starting replication. The row in mysql.ndb_apply_status with id 0 will be updated/inserted. | | | | --skip-table-check | Skip table structure check during restoring of data | 5.1.17 | | | --tab=path | Creates a tab-separated .txt file for each table in the given path | 5.1.18 | | | --verbose=# | Control level of verbosity in output | | | |
Typical options for this utility are shown here:
ndb_restore [-c connectstring ] -n node_id [-m] -b backup_id \
-r --backup_path=/path/to/backup/files
The -c option is used to specify a
connectstring which tells ndb_restore where
to locate the cluster management server. (See
Section 17.3.2.3, “The MySQL Cluster Connectstring”, for information
on connectstrings.) If this option is not used, then
ndb_restore attempts to connect to a
management server on localhost:1186 . This
utility acts as a cluster API node, and so requires a free
connection “slot” to connect to the cluster
management server. This means that there must be at least one
[api] or [mysqld]
section that can be used by it in the cluster
config.ini file. It is a good idea to
keep at least one empty [api] or
[mysqld] section in
config.ini that is not being used for a
MySQL server or other application for this reason (see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”).
You can verify that ndb_restore is
connected to the cluster by using the SHOW
command in the ndb_mgm management client.
You can also accomplish this from a system shell, as shown
here:
shell> ndb_mgm -e "SHOW"
-n is used to specify the node ID of the data
node on which the backups were taken.
The first time you run the ndb_restore
restoration program, you also need to restore the metadata. In
other words, you must re-create the database tables —
this can be done by running it with the -m
option. Note that the cluster should have an empty database
when starting to restore a backup. (In other words, you should
start ndbd with --initial
prior to performing the restore. You should also remove
manually any Disk Data files present in the data node's
DataDir .)
It is possible to restore data without restoring table
metadata. Prior to MySQL 5.1.17,
ndb_restore did not perform any checks of
table schemas; if a table was altered between the time the
backup was taken and when ndb_restore was
run, ndb_restore would still attempt to
restore the data to the altered table.
Beginning with MySQL 5.1.17, the default behavior is for
ndb_restore to fail with an error if table
data do not match the table schema; this can be overridden
using the --skip-table-check or
-s option. Prior to MySQL 5.1.21, if this
option is used, then ndb_restore attempts
to fit data into the existing table schema, but the result of
restoring a backup to a table schema that does not match the
original is unspecified.
Beginning with MySQL Cluster NDB 6.3.8,
ndb_restore supports limited
attribute promotion in much the same
way that it is supported by MySQL replication; that is, data
backed up from a column of a given type can generally be
restored to a column using a “larger, similar”
type. For example, data from a CHAR(20)
column can be restored to a column declared as
VARCHAR(20) ,
VARCHAR(30) , or
CHAR(30) ; data from a
MEDIUMINT column can be
restored to a column of type
INT or
BIGINT . See
Section 16.3.1.5.2, “Replication of Columns Having Different Data Types”,
for a table of type conversions currently supported by
attribute promotion.
Attribute promotion by ndb_restore must be
enabled explicitly, as follows:
Prepare the table to which the backup is to be restored.
ndb_restore cannot be used to
re-create the table with a different definition from the
original; this means that you must either create the
table manually, or alter the columns which you wish to
promote using ALTER TABLE
after restoring the table metadata but before restoring
the data.
Invoke ndb_restore with the
--promote-attributes option (short form
-A ) when restoring the table data.
Attribute promotion does not occur if this option is not
used; instead, the restore operation fails with an
error.
In addition to --promote-attributes , a
--preserve-trailing-spaces option is also
available for use with ndb_restore
beginning with MySQL Cluster NDB 6.3.8. This option (short
form -R ) causes trailing spaces to be
preserved when promoting a CHAR
column to VARCHAR or a
BINARY column to
VARBINARY . Otherwise, any
trailing spaces are dropped from column values when they are
inserted into the new columns.
The -b option is used to specify the ID or
sequence number of the backup, and is the same number shown by
the management client in the Backup
backup_id completed
message displayed upon completion of a backup. (See
Section 17.5.3.2, “Using The MySQL Cluster Management Client to Create a Backup”.)
Important
When restoring cluster backups, you must be sure to restore
all data nodes from backups having the same backup ID. Using
files from different backups will at best result in
restoring the cluster to an inconsistent state, and may fail
altogether.
--restore_epoch (short form:
-e ) adds (or restores) epoch information to
the cluster replication status table. This is useful for
starting replication on a MySQL Cluster replication slave.
When this option is used, the row in the
mysql.ndb_apply_status having
0 in the id column is
updated if it already exists; such a row is inserted if it
does not already exist. (See
Section 17.6.9, “MySQL Cluster Backups With MySQL Cluster Replication”.)
The path to the backup directory is required; this is supplied
to ndb_restore using the
--backup_path option, and must include the
subdirectory corresponding to the ID backup of the backup to
be restored. For example, if the data node's
DataDir is
/var/lib/mysql-cluster , then the backup
directory is
/var/lib/mysql-cluster/BACKUP , and the
backup files for the backup with the ID 3 can be found in
/var/lib/mysql-cluster/BACKUP/BACKUP-3 .
The path may be absolute or relative to the directory in which
the ndb_restore executable is located, and
may be optionally prefixed with backup_path= .
Note
Previous to MySQL 5.1.17 and MySQL Cluster NDB 6.1.5, the
path to the backup directory was specified as shown here,
with backup_path= being optional:
[backup_path=]/path/to/backup/files
Beginning with MySQL 5.1.17 and MySQL Cluster NDB 6.1.5,
this syntax changed to
--backup_path=/path/to/backup/files ,
to conform more closely with options used by other MySQL
programs; --backup_id is required, and
there is no short form for this option.
It is possible to restore a backup to a database with a
different configuration than it was created from. For example,
suppose that a backup with backup ID 12 ,
created in a cluster with two database nodes having the node
IDs 2 and 3 , is to be
restored to a cluster with four nodes. Then
ndb_restore must be run twice — once
for each database node in the cluster where the backup was
taken. However, ndb_restore cannot always
restore backups made from a cluster running one version of
MySQL to a cluster running a different MySQL version. See
Section 17.2.6.2, “MySQL Cluster 5.1 and MySQL Cluster NDB 6.x/7.x Upgrade and Downgrade
Compatibility”,
for more information.
Important
It is not possible to restore a backup made from a newer
version of MySQL Cluster using an older version of
ndb_restore. You can restore a backup
made from a newer version of MySQL to an older cluster, but
you must use a copy of ndb_restore from
the newer MySQL Cluster version to do so.
For example, to restore a cluster backup taken from a
cluster running MySQL Cluster NDB 6.2.15 to a cluster
running MySQL 5.1.20, you must use a copy of
ndb_restore from the MySQL Cluster NDB
6.2.15 distribution.
For more rapid restoration, the data may be restored in
parallel, provided that there is a sufficient number of
cluster connections available. That is, when restoring to
multiple nodes in parallel, you must have an
[api] or [mysqld]
section in the cluster config.ini file
available for each concurrent ndb_restore
process. However, the data files must always be applied before
the logs.
Formerly, when using ndb_restore to restore
a backup made from a MySQL 5.0 cluster to a 5.1 cluster,
VARCHAR columns were not
resized and were recreated using the 5.0 fixed format.
Beginning with MySQL 5.1.19, ndb_restore
recreates such VARCHAR columns
using MySQL Cluster 5.1's variable-width format. Also
beginning with MySQL 5.1.19, this behavior can be overridden
using the --no-upgrade option (short form:
-u ) when running
ndb_restore.
This option causes ndb_restore to print its
output to stdout . Beginning with MySQL
5.1.18, several additional options are available for use with
the --print_data option in generating data
dumps, either to stdout , or to a file.
These are similar to some of the options used with
mysqldump, and are shown in the following
list:
--tab , -T
This option causes
--print_data to create
dump files, one per table, each named
tbl_name .txt .
It requires as its argument the path to the directory
where the files should be saved; use .
for the current directory.
--fields-enclosed-by=string
Each column values are enclosed by the string passed to
this option (regardless of data type; see next item).
--fields-optionally-enclosed-by=string
The string passed to this option is used to enclose column
values containing character data (such as
CHAR ,
VARCHAR ,
BINARY ,
TEXT , or
ENUM ).
--fields-terminated-by=string
The string passed to this option is used to separate
column values. The default value is a tab character
(\t ).
--hex
If this option is used, all binary values are output in
hexadecimal format.
--fields-terminated-by=string
This option specifies the string used to end each line of
output. The default is a linefeed character
(\n ).
--append
When used with the
--tab and
--print_data options,
this causes the data to be appended to any existing files
having the same names.
Note
If a table has no explicit primary key, then the output
generated when using the
--print_data option
includes the table's hidden primary key.
Beginning with MySQL 5.1.18, it is possible to restore
selected databases, or to restore selected tables from a given
database using the syntax shown here:
ndb_restore other_options db_name ,[db_name [,...] | tbl_name [,tbl_name ][,...]]
In other words, you can specify either of the following to be
restored:
--include-databases=db_name [,db_name ][,...]
--include-tables=db_name.tbl_name [,db_name.tbl_name ][,...]
Beginning with MySQL Cluster NDB 6.3.22 and MySQL Cluster NDB
6.4.3, you can (and should) use instead the
--include-databases option or the
--include-tables option for restoring only
specific databases or tables, respectively.
--include-databases takes a comma-delimited
list of databases to be restored.
--include-tables takes a comma-delimited list
of tables (in
database .table
format) to be restored.
When --include-databases or
--include-tables is used, only those
databases or tables named by the option are restored; all
other databases and tables are excluded by
ndb_restore, and are not restored.
The following table shows several invocations of
ndb_restore using
--include-* options (other options possibly
required have been omitted for clarity), and the effects these
have on restoring from a MySQL Cluster backup:
Beginning with MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB
7.0.10, you can use these two options together. For example,
the following causes all tables in databases
db1 and db2 , together
with the tables t1 and
t2 in database db3 , to
be restored (and no other databases or tables):
shell> ndb_restore [...] --include-databases=db1,db2 --include-tables=db3.t1,db3.t2
(Again we have omitted other, possibly required, options in
the example just shown.)
Note
Prior to MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB
7.0.10, multiple --include-* options were
not handled correctly, and the result of the options shown
in the previous example was that only the tables
db3.t1 and db3.t2 were
actually restored. (Bug#48907)
--exclude-databases=db_name [,db_name ][,...]
--exclude-tables=db_name.tbl_name [,db_name.tbl_name ][,...]
Also beginning with MySQL Cluster NDB 6.3.22 and MySQL Cluster
NDB 6.4.3, it is possible to exclude from being restored one
or more databases or tables using the
ndb_restore options
--exclude-databases and
--exclude-tables .
--exclude-databases takes a comma-delimited
list of one or more databases which should not be restored.
--exclude-tables takes a comma-delimited list
of one or more tables (using
database .table
format) which should not be restored.
When --exclude-databases or
--exclude-tables is used, only those
databases or tables named by the option are excluded; all
other databases and tables are restored by
ndb_restore.
This table shows several invocations of
ndb_restore usng
--exclude-* options (other options possibly
required have been omitted for clarity), and the effects these
options have on restoring from a MySQL Cluster backup:
Beginning with MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB
7.0.10, you can use these two options together. For example,
the following causes all tables in all databases
except for databases
db1 and db2 , along with
the tables t1 and t2 in
database db3 , not to
be restored:
shell> ndb_restore [...] --exclude-databases=db1,db2 --exclude-tables=db3.t1,db3.t2
(Again, we have omitted other possibly necessary options in
the interest of clarity and brevity from the example just
shown.)
Note
Prior to MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB
7.0.10, multiple --exclude-* options were
not handled correctly, with the result that the options
shown in the previous example caused ndb_restore to exclude
only the tables db3.t1 and
db3.t2 . (Bug#48907)
Beginning with MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB
7.0.10, you can use --include-* and
--exclude-* options together, subject to the
following rules:
The actions of all --include-* and
--exclude-* options are cumulative.
All --include-* and
--exclude-* options are evaluated in the
order passed to ndb_restore, from right to left.
In the event of conflicting options, the first (rightmost)
option takes precedence. In other words, the first option
(going from right to left) that matches against a given
database or table “wins”.
For example, the following set of options causes
ndb_restore to restore all tables from
database db1 except
db1.t1 , while restoring no other tables
from any other databases:
--include-databases=db1 --exclude-tables=db1.t1
However, reversing the order of the options just given simply
causes all tables from database db1 to be
restored (including db1.t1 , but no tables
from any other database), because the
--include-dabases option, being farthest to
the right, is the first match against database
db1 and thus takes precedence over any
other option that matches db1 or any tables
in db1 :
--exclude-tables=db1.t1 --include-databases=db1
Note
Prior to MySQL Cluster NDB 6.3.29 and MySQL Cluster NDB
7.0.10, it was not possible to use
--include-databases or
--include-tables together with
--exclude-databases or
--exclude-tables , as these combinations
were evaluated inconsistently. (Bug#48907)
--exclude-missing-columns
Beginning with MySQL Cluster NDB 6.3.26 and MySQL Cluster NDB
7.0.7, it is also possible to restore only selected table
columns using the --exclude-missing-columns
option. When this option is used,
ndb_restore ignores any columns missing
from tables being restored as compared to the versions of
those tables found in the backup. This option applies to all
tables being restored. If you wish to apply this option only
to selected tables or databases, you can use it in combination
with one or more of the options described in the previous
paragraph to do so, then restore data to the remaining tables
using a complementary set of these options.
Error reporting.
ndb_restore reports both temporary and
permanent errors. In the case of temporary errors, it may
able to recover from them. Beginning with MySQL 5.1.12, it
reports Restore successful, but encountered
temporary error, please look at configuration in
such cases.
Important
After using ndb_restore to initialize a
MySQL Cluster for use in circular replication, binary logs
on the SQL node acting as the replication slave are not
automatically created, and you must cause them to be created
manually. In order to cause the binary logs to be created,
issue a SHOW TABLES statement
on that SQL node before running START
SLAVE .
This is a known issue with MySQL Cluster management, which
we intend to address in a future release.
17.4.18. ndb_select_all — Print Rows from an NDB Table
ndb_select_all prints all rows from an
NDB table to
stdout .
Usage:
ndb_select_all -c connect_string tbl_name -d db_name [> file_name ]
Additional Options:
--lock=lock_type ,
-l lock_type
Employs a lock when reading the table. Possible values for
lock_type are:
0 : Read lock
1 : Read lock with hold
2 : Exclusive read lock
There is no default value for this option.
--order=index_name ,
-o index_name
Orders the output according to the index named
index_name . Note that this is
the name of an index, not of a column, and that the index
must have been explicitly named when created.
--descending , -z
Sorts the output in descending order. This option can be
used only in conjunction with the -o
(--order ) option.
--header=FALSE
Excludes column headers from the output.
--useHexFormat -x
Causes all numeric values to be displayed in hexadecimal
format. This does not affect the output of numerals
contained in strings or datetime values.
--delimiter=character ,
-D character
Causes the character to be used
as a column delimiter. Only table data columns are
separated by this delimiter.
The default delimiter is the tab character.
--disk
Adds a disk reference column to the output. The column is
nonempty only for Disk Data tables having nonindexed
columns.
--rowid
Adds a ROWID column providing
information about the fragments in which rows are stored.
--gci
Adds a column to the output showing the global checkpoint
at which each row was last updated. See
Section 17.1, “MySQL Cluster Overview”, and
Section 17.5.4.2, “MySQL Cluster Log Events”, for more
information about checkpoints.
--tupscan , -t
Scan the table in the order of the tuples.
--nodata
Causes any table data to be omitted.
Sample Output:
Output from a MySQL SELECT
statement:
mysql> SELECT * FROM ctest1.fish;
+----+-----------+
| id | name |
+----+-----------+
| 3 | shark |
| 6 | puffer |
| 2 | tuna |
| 4 | manta ray |
| 5 | grouper |
| 1 | guppy |
+----+-----------+
6 rows in set (0.04 sec)
Output from the equivalent invocation of
ndb_select_all:
shell> ./ndb_select_all -c localhost fish -d ctest1
id name
3 [shark]
6 [puffer]
2 [tuna]
4 [manta ray]
5 [grouper]
1 [guppy]
6 rows returned
NDBT_ProgramExit: 0 - OK
Note that all string values are enclosed by square brackets
(“[ ...] ”)
in the output of ndb_select_all. For a
further example, consider the table created and populated as
shown here:
CREATE TABLE dogs (
id INT(11) NOT NULL AUTO_INCREMENT,
name VARCHAR(25) NOT NULL,
breed VARCHAR(50) NOT NULL,
PRIMARY KEY pk (id),
KEY ix (name)
)
TABLESPACE ts STORAGE DISK
ENGINE=NDBCLUSTER;
INSERT INTO dogs VALUES
('', 'Lassie', 'collie'),
('', 'Scooby-Doo', 'Great Dane'),
('', 'Rin-Tin-Tin', 'Alsatian'),
('', 'Rosscoe', 'Mutt');
This demonstrates the use of several additional
ndb_select_all options:
shell> ./ndb_select_all -d ctest1 dogs -o ix -z --gci --disk
GCI id name breed DISK_REF
834461 2 [Scooby-Doo] [Great Dane] [ m_file_no: 0 m_page: 98 m_page_idx: 0 ]
834878 4 [Rosscoe] [Mutt] [ m_file_no: 0 m_page: 98 m_page_idx: 16 ]
834463 3 [Rin-Tin-Tin] [Alsatian] [ m_file_no: 0 m_page: 34 m_page_idx: 0 ]
835657 1 [Lassie] [Collie] [ m_file_no: 0 m_page: 66 m_page_idx: 0 ]
4 rows returned
NDBT_ProgramExit: 0 - OK
17.4.19. ndb_select_count — Print Row Counts for NDB Tables
ndb_select_count prints the number of rows
in one or more NDB tables. With a
single table, the result is equivalent to that obtained by
using the MySQL statement SELECT COUNT(*) FROM
tbl_name .
Usage:
ndb_select_count [-c connect_string ] -ddb_name tbl_name [, tbl_name2 [, ...]]
Additional Options: None that
are specific to this application. However, you can obtain row
counts from multiple tables in the same database by listing
the table names separated by spaces when invoking this
command, as shown under Sample
Output.
Sample Output:
shell> ./ndb_select_count -c localhost -d ctest1 fish dogs
6 records in table fish
4 records in table dogs
NDBT_ProgramExit: 0 - OK
17.4.20. ndb_show_tables — Display List of NDB Tables
ndb_show_tables displays a list of all
NDB database objects in the
cluster. By default, this includes not only both user-created
tables and NDB system tables, but
NDB -specific indexes, internal
triggers, and MySQL Cluster Disk Data objects as well.
The following table includes options that are specific to the
MySQL Cluster program ndb_show_tables.
Additional descriptions follow the table. For options common
to all MySQL Cluster programs, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.15. ndb_show_tables Command Line
Options
Usage:
ndb_show_tables [-c connect_string ]
--database , -d
Specifies the name of the database in which the tables are
found.
--loops , -l
Specifies the number of times the utility should execute.
This is 1 when this option is not specified, but if you do
use the option, you must supply an integer argument for
it.
--parsable , -p
Using this option causes the output to be in a format
suitable for use with
LOAD DATA
INFILE .
--show-temp-status
If specified, this causes temporary tables to be
displayed.
--type , -t
Can be used to restrict the output to one type of object,
specified by an integer type code as shown here:
1: System table
2: User-created table
3: Unique hash index
Any other value causes all
NDB database objects to be
listed (the default).
--unqualified , -u
If specified, this causes unqualified object names to be
displayed.
17.4.21. ndb_size.pl — NDBCLUSTER Size Requirement Estimator
This is a Perl script that can be used to estimate the amount
of space that would be required by a MySQL database if it were
converted to use the NDBCLUSTER
storage engine. Unlike the other utilities discussed in this
section, it does not require access to a MySQL Cluster (in
fact, there is no reason for it to do so). However, it does
need to access the MySQL server on which the database to be
tested resides.
Requirements:
A running MySQL server. The server instance does not have
to provide support for MySQL Cluster.
A working installation of Perl.
The DBI module, which can be obtained
from CPAN if it is not already part of your Perl
installation. (Many Linux and other operating system
distributions provide their own packages for this
library.)
Previous to MySQL 5.1.18, ndb_size.pl
also required the HTML::Template module
and an associated template file
share/mysql/ndb_size.tmpl . Beginning
with MySQL 5.1.18, ndb_size.tmpl is
no longer needed (or included).
A MySQL user account having the necessary privileges. If
you do not wish to use an existing account, then creating
one using GRANT USAGE ON
db_name .* —
where db_name is the name of
the database to be examined — is sufficient for this
purpose.
ndb_size.pl can also be found in the
MySQL sources in storage/ndb/tools . If
this file is not present in your MySQL installation, you can
obtain it from the
MySQL
Forge project page.
The following table includes options that are specific to the
MySQL Cluster program ndb_size.pl.
Additional descriptions follow the table. For options common
to all MySQL Cluster programs, see
Section 17.4.23, “Options Common to MySQL Cluster Programs”.
Table 17.16. ndb_size.pl Command Line
Options
Usage:
perl ndb_size.pl db_name |ALL] [--hostname=host [:port ]] [--socket=socket ] [--user=user ] \
[--password=password ] [--help|-h] [--format=(html|text)] [--loadqueries=file_name ] [--savequeries=file_name ]
By default, this utility attempts to analyze all databases on
the server. You can specify a single database using the
--database option; the default behavior can
be made explicit by using ALL for the name
of the database. You can also exclude one or more databases by
using the --excludedbs with a comma-separated
list of the names of the databases to be skipped. Similarly,
you can cause specific tables to be skipped by listing their
names, separated by commas, following the optional
--excludetables option. A host name (and
possibly a port as well) can be specified using
--hostname ; the default is
localhost:3306 . If necessary, you can
specify a socket; the default is
/var/lib/mysql.sock . A MySQL user name
and password can be specified the corresponding options shown.
It also possible to control the format of the output using the
--format option; this can take either of the
values html or text ,
with text being the default. An example of
the text output is shown here:
shell> ndb_size.pl --database=test --socket=/tmp/mysql.sock
ndb_size.pl report for database: 'test' (1 tables)
--------------------------------------------------
Connected to: DBI:mysql:host=localhost;mysql_socket=/tmp/mysql.sock
Including information for versions: 4.1, 5.0, 5.1
test.t1
-------
DataMemory for Columns (* means varsized DataMemory):
Column Name Type Varsized Key 4.1 5.0 5.1
HIDDEN_NDB_PKEY bigint PRI 8 8 8
c2 varchar(50) Y 52 52 4*
c1 int(11) 4 4 4
-- -- --
Fixed Size Columns DM/Row 64 64 12
Varsize Columns DM/Row 0 0 4
DataMemory for Indexes:
Index Name Type 4.1 5.0 5.1
PRIMARY BTREE 16 16 16
-- -- --
Total Index DM/Row 16 16 16
IndexMemory for Indexes:
Index Name 4.1 5.0 5.1
PRIMARY 33 16 16
-- -- --
Indexes IM/Row 33 16 16
Summary (for THIS table):
4.1 5.0 5.1
Fixed Overhead DM/Row 12 12 16
NULL Bytes/Row 4 4 4
DataMemory/Row 96 96 48 (Includes overhead, bitmap and indexes)
Varsize Overhead DM/Row 0 0 8
Varsize NULL Bytes/Row 0 0 4
Avg Varside DM/Row 0 0 16
No. Rows 0 0 0
Rows/32kb DM Page 340 340 680
Fixedsize DataMemory (KB) 0 0 0
Rows/32kb Varsize DM Page 0 0 2040
Varsize DataMemory (KB) 0 0 0
Rows/8kb IM Page 248 512 512
IndexMemory (KB) 0 0 0
Parameter Minimum Requirements
------------------------------
* indicates greater than default
Parameter Default 4.1 5.0 5.1
DataMemory (KB) 81920 0 0 0
NoOfOrderedIndexes 128 1 1 1
NoOfTables 128 1 1 1
IndexMemory (KB) 18432 0 0 0
NoOfUniqueHashIndexes 64 0 0 0
NoOfAttributes 1000 3 3 3
NoOfTriggers 768 5 5 5
For debugging purposes, the Perl arrays containing the queries
run by this script can be read from the file specified using
can be saved to a file using --savequeries ; a
file containing such arrays to be read in during script
execution can be specified using
--loadqueries . Neither of these options has a
default value.
To produce output in HTML format, use the
--format option and redirect the output to a
file, as shown in this example:
shell> ndb_size.pl --database=test --socket=/tmp/mysql.sock --format=html > ndb_size.html
(Without the redirection, the output is sent to
stdout .) This figure shows a portion of the
generated ndb_size.html output file, as
viewed in a Web browser:
The output from this script includes:
Minimum values for the DataMemory ,
IndexMemory ,
MaxNoOfTables ,
MaxNoOfAttributes ,
MaxNoOfOrderedIndexes ,
MaxNoOfUniqueHashIndexes , and
MaxNoOfTriggers configuration
parameters required to accommodate the tables analyzed.
Memory requirements for all of the tables, attributes,
ordered indexes, and unique hash indexes defined in the
database.
The IndexMemory and
DataMemory required per table and table
row.
Note
Prior to MySQL 5.1.23, MySQL Cluster NDB 6.2.5, and MySQL
Cluster NDB 6.3.7, ndb_size.pl was
invoked as shown here:
perl ndb_size.pl db_name hostname username password > file_name .html
For more information about this change, see Bug#28683 and
Bug#28253.
17.4.22. ndb_waiter — Wait for MySQL Cluster to Reach a Given Status
ndb_waiter repeatedly (each 100
milliseconds) prints out the status of all cluster data nodes
until either the cluster reaches a given status or the
--timeout limit is exceeded, then exits. By
default, it waits for the cluster to achieve
STARTED status, in which all nodes have
started and connected to the cluster. This can be overridden
using the --no-contact and
--not-started options (see
Additional
Options).
The node states reported by this utility are as follows:
NO_CONTACT : The node cannot be
contacted.
UNKNOWN : The node can be contacted, but
its status is not yet known. Usually, this means that the
node has received a START or
RESTART command from the management
server, but has not yet acted on it.
NOT_STARTED : The node has stopped, but
remains in contact with the cluster. This is seen when
restarting the node using the management client's
RESTART command.
STARTING : The node's
ndbd process has started, but the node
has not yet joined the cluster.
STARTED : The node is operational, and
has joined the cluster.
SHUTTING_DOWN : The node is shutting
down.
SINGLE USER MODE : This is shown for all
cluster data nodes when the cluster is in single user
mode.
Usage:
ndb_waiter [-c connect_string ]
Additional Options:
--no-contact , -n
Instead of waiting for the STARTED
state, ndb_waiter continues running
until the cluster reaches NO_CONTACT
status before exiting.
--not-started
Instead of waiting for the STARTED
state, ndb_waiter continues running
until the cluster reaches NOT_STARTED
status before exiting.
--timeout=seconds ,
-t seconds
Time to wait. The program exits if the desired state is
not achieved within this number of seconds. The default is
120 seconds (1200 reporting cycles).
Sample Output.
Shown here is the output from ndb_waiter
when run against a 4-node cluster in which two nodes have
been shut down and then started again manually. Duplicate
reports (indicated by “... ”)
are omitted.
shell> ./ndb_waiter -c localhost
Connecting to mgmsrv at (localhost)
State node 1 STARTED
State node 2 NO_CONTACT
State node 3 STARTED
State node 4 NO_CONTACT
Waiting for cluster enter state STARTED
...
State node 1 STARTED
State node 2 UNKNOWN
State node 3 STARTED
State node 4 NO_CONTACT
Waiting for cluster enter state STARTED
...
State node 1 STARTED
State node 2 STARTING
State node 3 STARTED
State node 4 NO_CONTACT
Waiting for cluster enter state STARTED
...
State node 1 STARTED
State node 2 STARTING
State node 3 STARTED
State node 4 UNKNOWN
Waiting for cluster enter state STARTED
...
State node 1 STARTED
State node 2 STARTING
State node 3 STARTED
State node 4 STARTING
Waiting for cluster enter state STARTED
...
State node 1 STARTED
State node 2 STARTED
State node 3 STARTED
State node 4 STARTING
Waiting for cluster enter state STARTED
...
State node 1 STARTED
State node 2 STARTED
State node 3 STARTED
State node 4 STARTED
Waiting for cluster enter state STARTED
NDBT_ProgramExit: 0 - OK
Note
If no connectstring is specified, then
ndb_waiter tries to connect to a
management on localhost , and reports
Connecting to mgmsrv at (null) .
17.4.23. Options Common to MySQL Cluster Programs
All MySQL Cluster programs (except for mysqld)
take the options described in this section. Users of earlier MySQL
Cluster versions should note that some of these options have been
changed to make them consistent with one another as well as with
mysqld. You can use the --help
option with any MySQL Cluster program to view a list of the
options which it supports.
The options in the following table are common to all MySQL Cluster
executables.
Table 17.17. Common MySQL Cluster Command Line Options Format | Description | Introduction | Deprecated | Removed |
---|
--character-sets-dir=name | Directory where character sets are | | | | --ndb-connectstring=name | Set connect string for connecting to ndb_mgmd. Syntax: [nodeid=<id>;][host=]<hostname>[:<port>]. Overrides specifying entries in NDB_CONNECTSTRING and my.cnf | | | | --core-file | Write core on errors (defaults to TRUE in debug builds) | | | | --debug=options | Enable output from debug calls. Can be used only for versions compiled with debugging enabled | | | | --execute=name | Execute command and exit | | | | --help | Display help message and exit | | | | --ndb-mgmd-host=name | Set host and port for connecting to ndb_mgmd. Syntax: <hostname>[:<port>]. Overrides specifying entries in NDB_CONNECTSTRING and my.cnf | | | | --ndb-nodeid=# | Set node id for this node | | | | --ndb-optimized-node-selection | Select nodes for transactions in a more optimal way | | | | --ndb-shm | Allow optimizing using shared memory connections when available (EXPERIMENTAL ONLY) | | | | -V | Output version information and exit | | | |
For options specific to individual MySQL Cluster programs, see
Section 17.4, “MySQL Cluster Programs”.
See Section 17.3.4.2, “mysqld Command Options for MySQL Cluster”, for
mysqld options relating to MySQL Cluster.
--help --usage ,
-?
Prints a short list with descriptions of the available command
options.
--character-sets-dir=name
Tells the program where to find character set information.
--connect-string=connect_string ,
-c connect_string
connect_string sets the
connectstring to the management server as a command option.
shell> ndbd --connect-string="nodeid=2;host=ndb_mgmd.mysql.com:1186"
For more information, see
Section 17.3.2.3, “The MySQL Cluster Connectstring”.
--core-file
Write a core file if the program dies. The name and location
of the core file are system-dependent. (For MySQL Cluster
programs nodes running on Linux, the default location is the
program's working directory — for a data node, this
is the node's DataDir .) For some
systems, there may be restrictions or limitations; for
example, it might be necessary to execute ulimit -c
unlimited before starting the server. Consult your
system documentation for detailed information.
If MySQL Cluster was built using the --debug
option for configure, then
--core-file is enabled by default. For
regular builds, --core-file is disabled by
default.
--debug[=options ]
This option can be used only for versions compiled with
debugging enabled. It is used to enable output from debug
calls in the same manner as for the mysqld
process.
--execute=command ,
-e command
Can be used to send a command to a Cluster executable from the
system shell. For example, either of the following:
shell> ndb_mgm -e "SHOW"
or
shell> ndb_mgm --execute="SHOW"
is equivalent to
ndb_mgm> SHOW
This is analogous to how the --execute or
-e option works with the
mysql command-line client. See
Section 4.2.3.1, “Using Options on the Command Line”.
--ndb-mgmd-host=host [:port ]
Can be used to set the host and port number of the management
server to connect to.
--ndb-nodeid=#
Sets this node's MySQL Cluster node ID. The
range of permitted values depends on the type of the node
(data, management, or API) and the version of the MySQL
Cluster software which is running on it. See
Section 17.1.5.2, “Limits and Differences of MySQL Cluster from Standard MySQL Limits”, for more
information.
--ndb-optimized-node-selection
Optimize selection of nodes for transactions. Enabled by
default.
--version , -V
Prints the MySQL Cluster version number of the executable. The
version number is relevant because not all versions can be
used together, and the MySQL Cluster startup process verifies
that the versions of the binaries being used can co-exist in
the same cluster. This is also important when performing an
online (rolling) software upgrade or downgrade of MySQL
Cluster. (See
Section 17.2.6.1, “Performing a Rolling Restart of a MySQL Cluster”).
17.5. Management of MySQL Cluster
Managing a MySQL Cluster involves a number of tasks, the first of
which is to configure and start MySQL Cluster. This is covered in
Section 17.3, “MySQL Cluster Configuration”, and
Section 17.4, “MySQL Cluster Programs”.
The next few sections cover the management of a running MySQL
Cluster.
For information about security issues relating to management and
deployment of a MySQL Cluster, see
Section 17.5.9, “MySQL Cluster Security Issues”.
There are essentially two methods of actively managing a running
MySQL Cluster. The first of these is through the use of commands
entered into the management client whereby cluster status can be
checked, log levels changed, backups started and stopped, and nodes
stopped and started. The second method involves studying the
contents of the cluster log
ndb_node_id _cluster.log ;
this is usually found in the management server's
DataDir directory, but this location can be
overridden using the LogDestination option
— see Section 17.3.2.5, “Defining a MySQL Cluster Management Server”, for
details. (Recall that node_id represents
the unique identifier of the node whose activity is being logged.)
The cluster log contains event reports generated by
ndbd. It is also possible to send cluster log
entries to a Unix system log.
Some aspects of the cluster's operation can be also be monitored
from an SQL node using the
SHOW ENGINE NDB
STATUS statement. See Section 12.5.5.16, “SHOW ENGINE Syntax”, for
more information.
In MySQL Cluster NDB 7.1.1 and later, detailed information about
cluster operations is available in real time via an SQL interface
using the ndbinfo database. For more information,
see Section 17.5.8, “The ndbinfo MySQL Cluster Information Database”.
17.5.1. Summary of MySQL Cluster Start Phases
This section provides a simplified outline of the steps involved
when MySQL Cluster data nodes are started. More complete
information can be found in
MySQL Cluster Start Phases.
These phases are the same as those reported in the output from the
node_id STATUS
command in the management client (see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”). In MySQL
Cluster NDB 7.1.1 and later, these start phases are also reported
in the start_phase column of the
ndbinfo.nodes table (see
Section 17.5.8.7, “The ndbinfo nodes Table”).
Start types.
There are several different startup types and modes, as shown
here:
Initial Start.
The cluster starts with a clean file system on all data
nodes. This occurs either when the cluster started for
the very first time, or when all data nodes are
restarted using the --initial option.
Note
Disk Data files are not removed when restarting a
node using --initial .
System Restart.
The cluster starts and reads data stored in the data
nodes. This occurs when the cluster has been shut down
after having been in use, when it is desired for the
cluster to resume operations from the point where it
left off.
Node Restart.
This is the online restart of a cluster node while the
cluster itself is running.
Initial Node Restart.
This is the same as a node restart, except that the node
is reinitialized and started with a clean file system.
Setup and initialization (Phase -1).
Prior to startup, each data node (ndbd
process) must be initialized. Initialization consists of the
following steps:
Obtain a node ID
Fetch configuration data
Allocate ports to be used for inter-node communications
Allocate memory according to settings obtained from the
configuration file
When a data node or SQL node first connects to the management
node, it reserves a cluster node ID. To make sure that no other
node allocates the same node ID, this ID is retained until the
node has managed to connect to the cluster and at least one
ndbd reports that this node is connected. This
retention of the node ID is guarded by the connection between the
node in question and ndb_mgmd.
Normally, in the event of a problem with the node, the node
disconnects from the management server, the socket used for the
connection is closed, and the reserved node ID is freed. However,
if a node is disconnected abruptly — for example, due to a
hardware failure in one of the cluster hosts, or because of
network issues — the normal closing of the socket by the
operating system may not take place. In this case, the node ID
continues to be reserved and not released until a TCP timeout
occurs 10 or so minutes later.
To take care of this problem, you can use PURGE STALE
SESSIONS . Running this statement forces all reserved
node IDs to be checked; any that are not being used by nodes
actually connected to the cluster are then freed.
Beginning with MySQL 5.1.11, timeout handling of node ID
assignments is implemented. This performs the ID usage checks
automatically after approximately 20 seconds, so that
PURGE STALE SESSIONS should no longer be
necessary in a normal Cluster start.
After each data node has been initialized, the cluster startup
process can proceed. The stages which the cluster goes through
during this process are listed here:
Phase 0.
The NDBFS and
NDBCNTR blocks start (see
NDB Kernel Blocks). The
cluster file system is cleared, if the cluster was started
with the --initial option.
Phase 1.
In this stage, all remaining
NDB kernel blocks are
started. Cluster connections are set up, inter-block
communications are established, and Cluster heartbeats are
started. In the case of a node restart, API node
connections are also checked.
Note
When one or more nodes hang in Phase 1 while the
remaining node or nodes hang in Phase 2, this often
indicates network problems. One possible cause of such
issues is one or more cluster hosts having multiple
network interfaces. Another common source of problems
causing this condition is the blocking of TCP/IP ports
needed for communications between cluster nodes. In
the latter case, this is often due to a misconfigured
firewall.
Phase 2.
The NDBCNTR kernel block checks the
states of all existing nodes. The master node is chosen,
and the cluster schema file is initialized.
Phase 3.
The DBLQH and DBTC
kernel blocks set up communications between them. The
startup type is determined; if this is a restart, the
DBDIH block obtains permission to
perform the restart.
Phase 4.
For an initial start or initial node restart, the redo log
files are created. The number of these files is equal to
NoOfFragmentLogFiles .
For a system restart:
For a node restart, find the tail of the redo log.
Phase 5.
Most of the database-related portion of a data node start
is performed during this phase. For an initial start or
system restart, a local checkpoint is executed, followed
by a global checkpoint. Periodic checks of memory usage
begin during this phase, and any required node takeovers
are performed.
Phase 6.
In this phase, node groups are defined and set up.
Phase 7.
The arbitrator node is selected and begins to function.
The next backup ID is set, as is the backup disk write
speed. Nodes reaching this start phase are marked as
Started . It is now possible for API
nodes (including SQL nodes) to connect to the cluster.
connect.
Phase 8.
If this is a system restart, all indexes are rebuilt (by
DBDIH ).
Phase 9.
The node internal startup variables are reset.
Phase 100 (OBSOLETE).
Formerly, it was at this point during a node restart or
initial node restart that API nodes could connect to the
node and begin to receive events. Currently, this phase is
empty.
Phase 101.
At this point in a node restart or initial node restart,
event delivery is handed over to the node joining the
cluster. The newly-joined node takes over responsibility
for delivering its primary data to subscribers. This phase
is also referred to as SUMA
handover phase.
After this process is completed for an initial start or system
restart, transaction handling is enabled. For a node restart or
initial node restart, completion of the startup process means that
the node may now act as a transaction coordinator.
17.5.2. Commands in the MySQL Cluster Management Client
In addition to the central configuration file, a cluster may also
be controlled through a command-line interface available through
the management client ndb_mgm. This is the
primary administrative interface to a running cluster.
Commands for the event logs are given in
Section 17.5.4, “Event Reports Generated in MySQL Cluster”; commands for
creating backups and restoring from them are provided in
Section 17.5.3, “Online Backup of MySQL Cluster”.
The management client has the following basic commands. In the
listing that follows, node_id denotes
either a database node ID or the keyword ALL ,
which indicates that the command should be applied to all of the
cluster's data nodes.
HELP
Displays information on all available commands.
SHOW
Displays information on the cluster's status.
Note
In a cluster where multiple management nodes are in use,
this command displays information only for data nodes that
are actually connected to the current management server.
node_id START
Brings online the data node identified by
node_id (or all data nodes).
ALL START works on all data nodes only, and
does not affect management nodes.
Important
To use this command to bring a data node online, the data
node must have been started using ndbd
--nostart or ndbd -n.
node_id STOP
Stops the data or management node identified by
node_id . Note that ALL
STOP works to stop all data nodes only, and does not
affect management nodes.
A node affected by this command disconnects from the cluster,
and its associated ndbd or
ndb_mgmd process terminates.
node_id RESTART [-n] [-i]
[-a]
Restarts the data node identified by
node_id (or all data nodes).
Using the -i option with
RESTART causes the data node to perform an
initial restart; that is, the node's file system is deleted
and recreated. The effect is the same as that obtained from
stopping the data node process and then starting it again
using ndbd --initial from the system shell.
Note that backup files and Disk Data files are not removed
when this option is used.
Using the -n option causes the data node
process to be restarted, but the data node is not actually
brought online until the appropriate START
command is issued. The effect of this option is the same as
that obtained from stopping the data node and then starting it
again using ndbd --nostart or ndbd
-n from the system shell.
Using the -a causes all current transactions
relying on this node to be aborted. No GCP check is done when
the node rejoins the cluster.
node_id STATUS
Displays status information for the data node identified by
node_id (or for all data nodes).
node_id REPORT
report-type
Displays a report of type
report-type for the data node
identified by node_id , or for all
data nodes using ALL .
Currently, there are two accepted values for
report-type :
BackupStatus provides a status report
on a cluster backup in progress
MemoryUsage displays how much data
memory and index memory is being used by each data node
as shown in this example:
ndb_mgm> ALL REPORT MEMORY
Node 1: Data usage is 5%(177 32K pages of total 3200)
Node 1: Index usage is 0%(108 8K pages of total 12832)
Node 2: Data usage is 5%(177 32K pages of total 3200)
Node 2: Index usage is 0%(108 8K pages of total 12832)
For a data node running ndbmtd (MySQL
Cluster NDB 7.0 and later), this command reports index
usage on a per-thread basis; in other words, the report
for each ndbmtd data node shows an
Index usage ... entry for each
running thread.
In MySQL Cluster NDB 7.1.1 and later, this information
is also available from the
ndbinfo.memoryusage table; see
Section 17.5.8.6, “The ndbinfo memoryusage Table”.
The REPORT command was introduced in MySQL
Cluster NDB 6.2.3 and MySQL Cluster NDB 6.3.0.
ENTER SINGLE USER MODE
node_id
Enters single user mode, whereby only the MySQL server
identified by the node ID node_id
is allowed to access the database.
Important
Currently, it is not possible in for data nodes to join a
MySQL Cluster while it is running in single user mode.
(Bug#20395)
EXIT SINGLE USER MODE
Exits single user mode, allowing all SQL nodes (that is, all
running mysqld processes) to access the
database.
Note
It is possible to use EXIT SINGLE USER
MODE even when not in single user mode, although
the command has no effect in this case.
QUIT , EXIT
Terminates the management client.
This command does not affect any nodes connected to the
cluster.
SHUTDOWN
Shuts down all cluster data nodes and management nodes. To
exit the management client after this has been done, use
EXIT or QUIT .
This command does not shut down any SQL
nodes or API nodes that are connected to the cluster.
CREATE NODEGROUP nodeid [,
nodeid , ...]
Creates a new MySQL Cluster node group and causes data nodes
to join it.
This command is used after adding new data nodes online to a
MySQL Cluster, and causes them to join a new node group and
thus to begin participating fully in the cluster. The command
takes as its sole parameter a comma-separated list of node IDs
— these are the IDs of the nodes just added and started
that are to join the new node group. The number of nodes must
be the same as the number of nodes in each node group that is
already part of the cluster (each MySQL Cluster node group
must have the same number of nodes). In other words, if the
MySQL Cluster has 2 node groups of 2 data nodes each, then the
new node group must also have 2 data nodes.
The node group ID of the new node group created by this
command is determined automatically, and always the next
highest unused node group ID in the cluster; it is not
possible to set it manually.
This command was introduced in MySQL Cluster NDB 6.4.0. For
more information, see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
DROP NODEGROUP
nodegroup_id
Drops the MySQL Cluster node group with the given
nodegroup_id .
This command can be used to drop a node group from a MySQL
Cluster. DROP NODEGROUP takes as its sole
argument the node group ID of the node group to be dropped.
DROP NODEGROUP acts only to remove the data
nodes in the effected node group from that node group. It does
not stop data nodes, assign them to a different node group, or
remove them from the cluster's configuration. A data node
that does not belong to a node group is indicated in the
output of the management client
SHOW command with
no nodegroup in place of the node group ID,
like this (indicated using bold text):
id=3 @10.100.2.67 (5.1.41-ndb-7.0.12, no nodegroup)
Prior to MySQL Cluster NDB 7.0.4, the
SHOW output was not
updated correctly following DROP NODEGROUP .
(Bug#43413)
DROP NODEGROUP works only when all data
nodes in the node group to be dropped are completely empty of
any table data and table definitions. Since there is currently
no way using ndb_mgm or in the
mysql client to remove all data from a
specific data node or node group, this means that the command
succeeds only in the two following cases:
After issuing CREATE NODEGROUP in the
ndb_mgm client, but before issuing any
ALTER ONLINE
TABLE ... REORGANIZE PARTITION statements in the
mysql client.
After dropping all NDBCLUSTER
tables using DROP TABLE .
TRUNCATE TABLE does not
work for this purpose because this removes only the table
data; the data nodes continue to store an
NDBCLUSTER table's
definition until a DROP TABLE statement
is issued that causes the table metadata to be dropped.
DROP NODEGROUP was introduced in MySQL
Cluster NDB 6.4.0. For more information, see
Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
17.5.3. Online Backup of MySQL Cluster
The next few sections describe how to prepare for and then to
create a MySQL Cluster backup using the functionality for this
purpose found in the ndb_mgm management client.
To distinguish this type of backup from a backup made using
mysqldump, we sometimes refer to it as a
“native” MySQL Cluster backup. (For information about
the creation of backups with mysqldump, see
Section 4.5.4, “mysqldump — A Database Backup Program”.) Restoration of MySQL Cluster backups
is done using the ndb_restore utility provided
with the MySQL Cluster distribution; for information about
ndb_restore and its use in restoring MySQL
Cluster backups, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
17.5.3.1. MySQL Cluster Backup Concepts
A backup is a snapshot of the database at a given time. The
backup consists of three main parts:
Metadata.
The names and definitions of all database tables
Table records.
The data actually stored in the database tables at the
time that the backup was made
Transaction log.
A sequential record telling how and when data was stored
in the database
Each of these parts is saved on all nodes participating in the
backup. During backup, each node saves these three parts into
three files on disk:
BACKUP-backup_id .node_id .ctl
A control file containing control information and metadata.
Each node saves the same table definitions (for all tables
in the cluster) to its own version of this file.
BACKUP-backup_id -0.node_id .data
A data file containing the table records, which are saved on
a per-fragment basis. That is, different nodes save
different fragments during the backup. The file saved by
each node starts with a header that states the tables to
which the records belong. Following the list of records
there is a footer containing a checksum for all records.
BACKUP-backup_id .node_id .log
A log file containing records of committed transactions.
Only transactions on tables stored in the backup are stored
in the log. Nodes involved in the backup save different
records because different nodes host different database
fragments.
In the listing above, backup_id
stands for the backup identifier and
node_id is the unique identifier for
the node creating the file.
17.5.3.2. Using The MySQL Cluster Management Client to Create a Backup
Before starting a backup, make sure that the cluster is properly
configured for performing one. (See
Section 17.5.3.3, “Configuration for MySQL Cluster Backups”.)
The START BACKUP command is used to create a
backup:
START BACKUP [backup_id ] [wait_option ] [snapshot_option ]
wait_option :
WAIT {STARTED | COMPLETED} | NOWAIT
snapshot_option :
SNAPSHOTSTART | SNAPSHOTEND
Successive backups are automatically identified sequentially, so
the backup_id , an integer greater
than or equal to 1, is optional; if it is omitted, the next
available value is used. If an existing
backup_id value is used, the backup
fails with the error Backup failed: file already
exists. If used, the
backup_id must follow START
BACKUP immediately, before any other options are used.
Prior to MySQL Cluster NDB 6.2.17, 6.3.23, and 6.4.3, backup IDs
greater than 2147483648 (231) were
not supported correctly. (Bug#43042) Beginning with these
versions, the maximum possible backup ID is 4294967296
(232).
Note
Prior to MySQL Cluster NDB 7.0.5, when starting a backup using
ndb_mgm -e "START BACKUP", the
backup_id was required. (Bug#31754)
The wait_option can be used to
determine when control is returned to the management client
after a START BACKUP command is issued, as
shown in the following list:
If NOWAIT is specified, the management
client displays a prompt immediately, as seen here:
ndb_mgm> START BACKUP NOWAIT
ndb_mgm>
In this case, the management client can be used even while
it prints progress information from the backup process.
With WAIT STARTED the management client
waits until the backup has started before returning control
to the user, as shown here:
ndb_mgm> START BACKUP WAIT STARTED
Waiting for started, this may take several minutes
Node 2: Backup 3 started from node 1
ndb_mgm>
WAIT COMPLETED causes the management
client to wait until the backup process is complete before
returning control to the user.
WAIT COMPLETED is the default.
Beginning with MySQL Cluster NDB 6.4.0, a
snapshot_option can be used to
determine whether the backup matches the state of the cluster
when START BACKUP was issued, or when it was
completed. SNAPSHOTSTART causes the backup to
match the state of the cluster when the backup began;
SNAPSHOTEND causes the backup to reflect the
state of the cluster when the backup was finished.
SNAPSHOTEND is the default, and matches the
behavior found in previous MySQL Cluster releases.
Note
If you use the SNAPSHOTSTART option with
START BACKUP , and the
CompressedBackup parameter is enabled, only
the data and control files are compressed — the log file
is not compressed.
If both a wait_option and a
snapshot_option are used, they may be
specified in either order. For example, all of the following
commands are valid, assuming that there is no existing backup
having 4 as its ID:
START BACKUP WAIT STARTED SNAPSHOTSTART
START BACKUP SNAPSHOTSTART WAIT STARTED
START BACKUP 4 WAIT COMPLETED SNAPSHOTSTART
START BACKUP SNAPSHOTEND WAIT COMPLETED
START BACKUP 4 NOWAIT SNAPSHOTSTART
The procedure for creating a backup consists of the following
steps:
Start the management client (ndb_mgm),
if it not running already.
Execute the START BACKUP command.
This produces several lines of output indicating the
progress of the backup, as shown here:
ndb_mgm> START BACKUP
Waiting for completed, this may take several minutes
Node 2: Backup 1 started from node 1
Node 2: Backup 1 started from node 1 completed
StartGCP: 177 StopGCP: 180
#Records: 7362 #LogRecords: 0
Data: 453648 bytes Log: 0 bytes
ndb_mgm>
When the backup has started the management client displays
this message:
Backup backup_id started from node node_id
backup_id is the unique
identifier for this particular backup. This identifier is
saved in the cluster log, if it has not been configured
otherwise. node_id is the
identifier of the management server that is coordinating
the backup with the data nodes. At this point in the
backup process the cluster has received and processed the
backup request. It does not mean that the backup has
finished. An example of this statement is shown here:
Node 2: Backup 1 started from node 1
The management client indicates with a message like this
one that the backup has started:
Backup backup_id started from node node_id completed
As is the case for the notification that the backup has
started, backup_id is the
unique identifier for this particular backup, and
node_id is the node ID of the
management server that is coordinating the backup with the
data nodes. This output is accompanied by additional
information including relevant global checkpoints, the
number of records backed up, and the size of the data, as
shown here:
Node 2: Backup 1 started from node 1 completed
StartGCP: 177 StopGCP: 180
#Records: 7362 #LogRecords: 0
Data: 453648 bytes Log: 0 bytes
It is also possible to perform a backup from the system shell by
invoking ndb_mgm with the -e
or --execute option, as shown in this example:
shell> ndb_mgm -e "START BACKUP 6 WAIT COMPLETED SNAPSHOTSTART"
When using START BACKUP in this way, you must
specify the backup ID.
Cluster backups are created by default in the
BACKUP subdirectory of the
DataDir on each data node. This can be
overridden for one or more data nodes individually, or for all
cluster data nodes in the config.ini file
using the BackupDataDir configuration
parameter as discussed in
Identifying
Data Nodes. The backup files created for a backup with a
given backup_id are stored in a
subdirectory named
BACKUP-backup_id
in the backup directory.
To abort a backup that is already in progress:
Start the management client.
Execute this command:
ndb_mgm> ABORT BACKUP backup_id
The number backup_id is the
identifier of the backup that was included in the response
of the management client when the backup was started (in the
message Backup backup_id
started from node
management_node_id ).
The management client will acknowledge the abort request
with Abort of backup
backup_id ordered .
Note
At this point, the management client has not yet received
a response from the cluster data nodes to this request,
and the backup has not yet actually been aborted.
After the backup has been aborted, the management client
will report this fact in a manner similar to what is shown
here:
Node 1: Backup 3 started from 5 has been aborted. Error: 1321 - Backup aborted by user request: Permanent error: User defined error
Node 3: Backup 3 started from 5 has been aborted. Error: 1323 - 1323: Permanent error: Internal error
Node 2: Backup 3 started from 5 has been aborted. Error: 1323 - 1323: Permanent error: Internal error
Node 4: Backup 3 started from 5 has been aborted. Error: 1323 - 1323: Permanent error: Internal error
In this example, we have shown sample output for a cluster
with 4 data nodes, where the sequence number of the backup
to be aborted is 3 , and the management
node to which the cluster management client is connected has
the node ID 5 . The first node to complete
its part in aborting the backup reports that the reason for
the abort was due to a request by the user. (The remaining
nodes report that the backup was aborted due to an
unspecified internal error.)
Note
There is no guarantee that the cluster nodes respond to an
ABORT BACKUP command in any particular
order.
The Backup backup_id
started from node
management_node_id has been
aborted messages mean that the backup has been
terminated and that all files relating to this backup have
been removed from the cluster file system.
It is also possible to abort a backup in progress from a system
shell using this command:
shell> ndb_mgm -e "ABORT BACKUP backup_id "
Note
If there is no backup having the ID
backup_id running when an
ABORT BACKUP is issued, the management
client makes no response, nor is it indicated in the cluster
log that an invalid abort command was sent.
17.5.3.3. Configuration for MySQL Cluster Backups
Five configuration parameters are essential for backup:
BackupDataBufferSize
The amount of memory used to buffer data before it is
written to disk.
BackupLogBufferSize
The amount of memory used to buffer log records before these
are written to disk.
BackupMemory
The total memory allocated in a database node for backups.
This should be the sum of the memory allocated for the
backup data buffer and the backup log buffer.
BackupWriteSize
The default size of blocks written to disk. This applies for
both the backup data buffer and the backup log buffer.
BackupMaxWriteSize
The maximum size of blocks written to disk. This applies for
both the backup data buffer and the backup log buffer.
More detailed information about these parameters can be found in
Backup
Parameters.
17.5.3.4. MySQL Cluster Backup Troubleshooting
If an error code is returned when issuing a backup request, the
most likely cause is insufficient memory or disk space. You
should check that there is enough memory allocated for the
backup.
Important
If you have set BackupDataBufferSize and
BackupLogBufferSize and their sum is
greater than 4MB, then you must also set
BackupMemory as well. See
BackupMemory .
You should also make sure that there is sufficient space on the
hard drive partition of the backup target.
NDB does not support repeatable
reads, which can cause problems with the restoration process.
Although the backup process is “hot”, restoring a
MySQL Cluster from backup is not a 100% “hot”
process. This is due to the fact that, for the duration of the
restore process, running transactions get nonrepeatable reads
from the restored data. This means that the state of the data is
inconsistent while the restore is in progress.
MySQL Enterprise
MySQL Enterprise subscribers will find more information about
Cluster backup in the Knowledge Base article,
How Do I
Backup my Cluster Database. Access to the MySQL
Knowledge Base collection of articles is one of the advantages
of subscribing to MySQL Enterprise. For more information, see
http://www.mysql.com/products/enterprise/advisors.html.
17.5.4. Event Reports Generated in MySQL Cluster
In this section, we discuss the types of event logs provided by
MySQL Cluster, and the types of events that are logged.
MySQL Cluster provides two types of event log:
The cluster log, which includes events
generated by all cluster nodes. The cluster log is the log
recommended for most uses because it provides logging
information for an entire cluster in a single location.
By default, the cluster log is saved to a file named
ndb_node_id _cluster.log ,
(where node_id is the node ID of
the management server) in the same directory where the
ndb_mgm binary resides.
Cluster logging information can also be sent to
stdout or a syslog
facility in addition to or instead of being saved to a file,
as determined by the values set for the
DataDir and
LogDestination configuration parameters.
See Section 17.3.2.5, “Defining a MySQL Cluster Management Server”, for more
information about these parameters.
Node logs are local to each node.
Output generated by node event logging is written to the file
ndb_node_id _out.log
(where node_id is the node's node
ID) in the node's DataDir . Node event logs
are generated for both management nodes and data nodes.
Node logs are intended to be used only during application
development, or for debugging application code.
Both types of event logs can be set to log different subsets of
events.
Each reportable event can be distinguished according to three
different criteria:
Category: This can be any one of the
following values: STARTUP ,
SHUTDOWN , STATISTICS ,
CHECKPOINT , NODERESTART ,
CONNECTION , ERROR , or
INFO .
Priority: This is represented by one of
the numbers from 1 to 15 inclusive, where 1 indicates
“most important” and 15 “least
important.”
Severity Level: This can be any one of
the following values: ALERT ,
CRITICAL , ERROR ,
WARNING , INFO , or
DEBUG .
Both the cluster log and the node log can be filtered on these
properties.
The format used in the cluster log is as shown here:
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 1: Data usage is 2%(60 32K pages of total 2560)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 1: Index usage is 1%(24 8K pages of total 2336)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 1: Resource 0 min: 0 max: 639 curr: 0
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 2: Data usage is 2%(76 32K pages of total 2560)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 2: Index usage is 1%(24 8K pages of total 2336)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 2: Resource 0 min: 0 max: 639 curr: 0
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 3: Data usage is 2%(58 32K pages of total 2560)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 3: Index usage is 1%(25 8K pages of total 2336)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 3: Resource 0 min: 0 max: 639 curr: 0
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 4: Data usage is 2%(74 32K pages of total 2560)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 4: Index usage is 1%(25 8K pages of total 2336)
2007-01-26 19:35:55 [MgmSrvr] INFO -- Node 4: Resource 0 min: 0 max: 639 curr: 0
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 4: Node 9 Connected
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 1: Node 9 Connected
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 1: Node 9: API 5.1.41-ndb-7.0.12
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 2: Node 9 Connected
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 2: Node 9: API 5.1.41-ndb-7.0.12
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 3: Node 9 Connected
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 3: Node 9: API 5.1.41-ndb-7.0.12
2007-01-26 19:39:42 [MgmSrvr] INFO -- Node 4: Node 9: API 5.1.41-ndb-7.0.12
2007-01-26 19:59:22 [MgmSrvr] ALERT -- Node 2: Node 7 Disconnected
2007-01-26 19:59:22 [MgmSrvr] ALERT -- Node 2: Node 7 Disconnected
Each line in the cluster log contains the following information:
A timestamp in
YYYY -MM -DD
HH :MM :SS
format.
The type of node which is performing the logging. In the
cluster log, this is always [MgmSrvr] .
The severity of the event.
The ID of the node reporting the event.
A description of the event. The most common types of events to
appear in the log are connections and disconnections between
different nodes in the cluster, and when checkpoints occur. In
some cases, the description may contain status information.
17.5.4.1. MySQL Cluster Logging Management Commands
The following management commands are related to the cluster
log:
CLUSTERLOG ON
Turns the cluster log on.
CLUSTERLOG OFF
Turns the cluster log off.
CLUSTERLOG INFO
Provides information about cluster log settings.
node_id CLUSTERLOG
category =threshold
Logs category events with
priority less than or equal to
threshold in the cluster log.
CLUSTERLOG FILTER
severity_level
Toggles cluster logging of events of the specified
severity_level .
The following table describes the default setting (for all data
nodes) of the cluster log category threshold. If an event has a
priority with a value lower than or equal to the priority
threshold, it is reported in the cluster log.
Note that events are reported per data node, and that the
threshold can be set to different values on different nodes.
The STATISTICS category can provide a great
deal of useful data. See
Section 17.5.4.3, “Using CLUSTERLOG STATISTICS in the MySQL Cluster
Management Client”, for more
information.
Thresholds are used to filter events within each category. For
example, a STARTUP event with a priority of 3
is not logged unless the threshold for
STARTUP is set to 3 or higher. Only events
with priority 3 or lower are sent if the threshold is 3.
The following table shows the event severity levels.
Note
These correspond to Unix syslog levels,
except for LOG_EMERG and
LOG_NOTICE , which are not used or mapped.
Event severity levels can be turned on or off (using
CLUSTERLOG FILTER — see above). If a
severity level is turned on, then all events with a priority
less than or equal to the category thresholds are logged. If the
severity level is turned off then no events belonging to that
severity level are logged.
Important
Cluster log levels are set on a per
ndb_mgmd, per subscriber basis. This means
that, in a MySQL Cluster with multiple management servers,
using a CLUSTERLOG command in an instance
of ndb_mgm connected to one management
server affects only logs generated by that management server
but not by any of the others. This also means that, should one
of the management servers be restarted, only logs generated by
that management server are affected by the resetting of log
levels caused by the restart.
17.5.4.2. MySQL Cluster Log Events
An event report reported in the event logs has the following
format:
datetime [string ] severity -- message
For example:
09:19:30 2005-07-24 [NDB] INFO -- Node 4 Start phase 4 completed
This section discusses all reportable events, ordered by
category and severity level within each category.
In the event descriptions, GCP and LCP mean “Global
Checkpoint” and “Local Checkpoint”,
respectively.
CONNECTION
Events
These events are associated with connections between Cluster
nodes.
CHECKPOINT
Events
The logging messages shown here are associated with checkpoints.
STARTUP
Events
The following events are generated in response to the startup of
a node or of the cluster and of its success or failure. They
also provide information relating to the progress of the startup
process, including information concerning logging activities.
NODERESTART
Events
The following events are generated when restarting a node and
relate to the success or failure of the node restart process.
STATISTICS
Events
The following events are of a statistical nature. They provide
information such as numbers of transactions and other
operations, amount of data sent or received by individual nodes,
and memory usage.
ERROR Events
These events relate to Cluster errors and warnings. The presence
of one or more of these generally indicates that a major
malfunction or failure has occurred.
INFO Events
These events provide general information about the state of the
cluster and activities associated with Cluster maintenance, such
as logging and heartbeat transmission.
17.5.4.3. Using CLUSTERLOG STATISTICS in the MySQL Cluster
Management Client
The NDB management client's
CLUSTERLOG STATISTICS command can provide a
number of useful statistics in its output. Counters providing
information about the state of the cluster are updated at
5-second reporting intervals by the transaction coordinator (TC)
and the local query handler (LQH), and written to the cluster
log.
Transaction coordinator statistics.
Each transaction has one transaction coordinator, which is
chosen by one of the following methods:
Note
You can determine which TC selection method is used for
transactions started from a given SQL node using the
ndb_optimized_node_selection
system variable.
All operations within the same transaction use the same
transaction coordinator, which reports the following
statistics:
Trans count .
This is the number transactions started in the last
interval using this TC as the transaction coordinator.
Any of these transactions may have committed, have
been aborted, or remain uncommitted at the end of the
reporting interval.
Note
Transactions do not migrate between TCs.
Commit count .
This is the number of transactions using this TC as
the transaction coordinator that were committed in the
last reporting interval. Because some transactions
committed in this reporting interval may have started
in a previous reporting interval, it is possible for
Commit count to be greater than
Trans count .
Read count .
This is the number of primary key read operations
using this TC as the transaction coordinator that were
started in the last reporting interval, including
simple reads. This count also includes reads performed
as part of unique index operations. A unique index
read operation generates 2 primary key read operations
— 1 for the hidden unique index table, and 1 for
the table on which the read takes place.
Simple read count .
This is the number of simple read operations using
this TC as the transaction coordinator that were
started in the last reporting interval. This is a
subset of Read count . Because the
value of Simple read count is
incremented at a different point in time from
Read count , it can lag behind
Read count slightly, so it is
conceivable that Simple read count
is not equal to Read count for a
given reporting interval, even if all reads made
during that time were in fact simple reads.
Write count .
This is the number of primary key write operations
using this TC as the transaction coordinator that were
started in the last reporting interval. This includes
all inserts, updates, writes and deletes, as well as
writes performed as part of unique index operations.
Note
A unique index update operation can generate
multiple PK read and write operations on the index
table and on the base table.
AttrInfoCount .
This is the number of 32-bit data words received in
the last reporting interval for primary key operations
using this TC as the transaction coordinator. For
reads, this is proportional to the number of columns
requested. For inserts and updates, this is
proportional to the number of columns written, and the
size of their data. For delete operations, this is
usually zero. Unique index operations generate
multiple PK operations and so increase this count.
However, data words sent to describe the PK operation
itself, and the key information sent, are
not counted here. Attribute
information sent to describe columns to read for
scans, or to describe ScanFilters, is also not counted
in AttrInfoCount .
Concurrent Operations .
This is the number of primary key or scan operations
using this TC as the transaction coordinator that were
started during the last reporting interval but that
were not completed. Operations increment this counter
when they are started and decrement it when they are
completed; this occurs after the transaction commits.
Dirty reads and writes — as well as failed
operations — decrement this counter. The maximum
value that Concurrent Operations
can have is the maximum number of operations that a TC
block can support; currently, this is (2 *
MaxNoOfConcurrentOperations) + 16 +
MaxNoOfConcurrentTransactions . (For more
information about these configuration parameters, see
the Transaction Parameters
section of
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.)
Abort count .
This is the number of transactions using this TC as
the transaction coordinator that were aborted during
the last reporting interval. Because some transactions
that were aborted in the last reporting interval may
have started in a previous reporting interval,
Abort count can sometimes be
greater than Trans count .
Scans .
This is the number of table scans using this TC as the
transaction coordinator that were started during the
last reporting interval. This does not include range
scans (that is, ordered index scans).
Range scans .
This is the number of ordered index scans using this
TC as the transaction coordinator that were started in
the last reporting interval.
Local query handler statistics (Operations ).
There is 1 cluster event per local query handler block (that
is, 1 per data node process). Operations are recorded in the
LQH where the data they are operating on resides.
Note
A single transaction may operate on data stored in
multiple LQH blocks.
The Operations statistic provides the
number of local operations performed by this LQH block in the
last reporting interval, and includes all types of read and
write operations (insert, update, write, and delete
operations). This also includes operations used to replicate
writes — for example, in a 2-replica cluster, the write
to the primary replica is recorded in the primary LQH, and the
write to the backup will be recorded in the backup LQH. Unique
key operations may result in multiple local operations;
however, this does not include local
operations generated as a result of a table scan or ordered
index scan, which are not counted.
Process scheduler statistics.
In addition to the statistics reported by the transaction
coordinator and local query handler, each
ndbd process has a scheduler which also
provides useful metrics relating to the performance of a MySQL
Cluster. This scheduler runs in an infinite loop; during each
loop the scheduler performs the following tasks:
Read any incoming messages from sockets into a job
buffer.
Check whether there are any timed messages to be
executed; if so, put these into the job buffer as well.
Execute (in a loop) any messages in the job buffer.
Send any distributed messages that were generated by
executing the messages in the job buffer.
Wait for any new incoming messages.
Process scheduler statistics include the following:
Mean Loop Counter .
This is the number of loops executed in the third step
from the preceding list. This statistic increases in
size as the utilization of the TCP/IP buffer improves.
You can use this to monitor changes in performance as
you add new data node processes.
Mean send size and Mean receive
size .
These statistics allow you to gauge the efficiency of,
respectively writes and reads between nodes. The
values are given in bytes. Higher values mean a lower
cost per byte sent or received; the maximum value is
64K.
To cause all cluster log statistics to be logged, you can use
the following command in the NDB
management client:
ndb_mgm> ALL CLUSTERLOG STATISTICS=15
Note
Setting the threshold for STATISTICS to
15 causes the cluster log to become very verbose, and to
grow quite rapidly in size, in direct proportion to the
number of cluster nodes and the amount of activity in the
MySQL Cluster.
For more information about MySQL Cluster management client
commands relating to logging and reporting, see
Section 17.5.4.1, “MySQL Cluster Logging Management Commands”.
17.5.5. MySQL Cluster Log Messages
This section contains information about the messages written to
the cluster log in response to different cluster log events. It
provides additional, more specific information on
NDB transporter errors.
17.5.5.2. MySQL Cluster — NDB Transporter
Errors
This section lists error codes, names, and messages that are
written to the cluster log in the event of transporter errors.
17.5.6. MySQL Cluster Single User Mode
Single user mode allows the database
administrator to restrict access to the database system to a
single API node, such as a MySQL server (SQL node) or an instance
of ndb_restore. When entering single user mode,
connections to all other API nodes are closed gracefully and all
running transactions are aborted. No new transactions are
permitted to start.
Once the cluster has entered single user mode, only the designated
API node is granted access to the database.
You can use the ALL STATUS command in the
ndb_mgm client to see when the cluster has
entered single user mode. Beginning with MySQL Cluster NDB 7.1.1,
you can also check the status column of the
ndbinfo.nodes table (see
Section 17.5.8.7, “The ndbinfo nodes Table”, for more
information).
Example:
ndb_mgm> ENTER SINGLE USER MODE 5
After this command has executed and the cluster has entered single
user mode, the API node whose node ID is 5
becomes the cluster's only permitted user.
The node specified in the preceding command must be an API node;
attempting to specify any other type of node will be rejected.
Note
When the preceding command is invoked, all transactions running
on the designated node are aborted, the connection is closed,
and the server must be restarted.
The command EXIT SINGLE USER MODE changes the
state of the cluster's data nodes from single user mode to normal
mode. API nodes — such as MySQL Servers — waiting for
a connection (that is, waiting for the cluster to become ready and
available), are again permitted to connect. The API node denoted
as the single-user node continues to run (if still connected)
during and after the state change.
Example:
ndb_mgm> EXIT SINGLE USER MODE
There are two recommended ways to handle a node failure when
running in single user mode:
17.5.7. Quick Reference: MySQL Cluster SQL Statements
This section discusses several SQL statements that can prove
useful in managing and monitoring a MySQL server that is connected
to a MySQL Cluster, and in some cases provide information about
the cluster itself.
SHOW ENGINE NDB
STATUS ,
SHOW ENGINE
NDBCLUSTER STATUS
The output of this statement contains information about the
server's connection to the cluster, creation and usage of
MySQL Cluster objects, and binary logging for MySQL Cluster
replication.
See Section 12.5.5.16, “SHOW ENGINE Syntax”, for a usage example and
more detailed information.
SHOW ENGINES
This statement can be used to determine whether or not
clustering support is enabled in the MySQL server, and if so,
whether it is active.
See Section 12.5.5.17, “SHOW ENGINES Syntax”, for more detailed
information.
Note
In MySQL 5.1 and later, this statement does not support a
LIKE clause. However, you can
use LIKE to filter queries
against the
INFORMATION_SCHEMA.ENGINES , as
discussed in the next item.
SELECT * FROM INFORMATION_SCHEMA.ENGINES [WHERE
ENGINE LIKE 'NDB%']
This is the equivalent of SHOW
ENGINES , but uses the
ENGINES table of the
INFORMATION_SCHEMA database (available
beginning with MySQL 5.1.5). Unlike the case with the
SHOW ENGINES statement, it is
possible to filter the results using a
LIKE clause, and to select
specific columns to obtain information that may be of use in
scripts. For example, the following query shows whether the
server was built with NDB support
and, if so, whether it is enabled:
mysql> SELECT SUPPORT FROM INFORMATION_SCHEMA.ENGINES
-> WHERE ENGINE LIKE 'NDB%';
+---------+
| support |
+---------+
| ENABLED |
+---------+
See Section 20.18, “The INFORMATION_SCHEMA ENGINES Table”, for more information.
SHOW VARIABLES LIKE 'NDB%'
This statement provides a list of most server system variables
relating to the NDB storage
engine, and their values, as shown here:
mysql> SHOW VARIABLES LIKE 'NDB%';
+-------------------------------------+-------+
| Variable_name | Value |
+-------------------------------------+-------+
| ndb_autoincrement_prefetch_sz | 32 |
| ndb_cache_check_time | 0 |
| ndb_extra_logging | 0 |
| ndb_force_send | ON |
| ndb_index_stat_cache_entries | 32 |
| ndb_index_stat_enable | OFF |
| ndb_index_stat_update_freq | 20 |
| ndb_report_thresh_binlog_epoch_slip | 3 |
| ndb_report_thresh_binlog_mem_usage | 10 |
| ndb_use_copying_alter_table | OFF |
| ndb_use_exact_count | ON |
| ndb_use_transactions | ON |
+-------------------------------------+-------+
See Section 5.1.4, “Server System Variables”, for more
information.
SELECT * FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES
WHERE VARIABLE_NAME LIKE 'NDB%';
This statement is the equivalent of the
SHOW command described
in the previous item, and provides almost identical output, as
shown here:
mysql> SELECT * FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES
-> WHERE VARIABLE_NAME LIKE 'NDB%';
+-------------------------------------+----------------+
| VARIABLE_NAME | VARIABLE_VALUE |
+-------------------------------------+----------------+
| NDB_AUTOINCREMENT_PREFETCH_SZ | 32 |
| NDB_CACHE_CHECK_TIME | 0 |
| NDB_EXTRA_LOGGING | 0 |
| NDB_FORCE_SEND | ON |
| NDB_INDEX_STAT_CACHE_ENTRIES | 32 |
| NDB_INDEX_STAT_ENABLE | OFF |
| NDB_INDEX_STAT_UPDATE_FREQ | 20 |
| NDB_REPORT_THRESH_BINLOG_EPOCH_SLIP | 3 |
| NDB_REPORT_THRESH_BINLOG_MEM_USAGE | 10 |
| NDB_USE_COPYING_ALTER_TABLE | OFF |
| NDB_USE_EXACT_COUNT | ON |
| NDB_USE_TRANSACTIONS | ON |
+-------------------------------------+----------------+
Unlike the case with the
SHOW command, it is
possible to select individual columns. For example:
mysql> SELECT VARIABLE_VALUE
-> FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES
-> WHERE VARIABLE_NAME = 'ndb_force_send';
+----------------+
| VARIABLE_VALUE |
+----------------+
| ON |
+----------------+
See Section 20.25, “The INFORMATION_SCHEMA GLOBAL_VARIABLES and
SESSION_VARIABLES
Tables”, and
Section 5.1.4, “Server System Variables”, for more
information.
SHOW STATUS LIKE 'NDB%'
This statement shows at a glance whether or not the MySQL
server is acting as a cluster SQL node, and if so, it provides
the MySQL server's cluster node ID, the host name and port for
the cluster management server to which it is connected, and
the number of data nodes in the cluster, as shown here:
mysql> SHOW STATUS LIKE 'NDB%';
+--------------------------+---------------+
| Variable_name | Value |
+--------------------------+---------------+
| Ndb_cluster_node_id | 10 |
| Ndb_config_from_host | 192.168.0.103 |
| Ndb_config_from_port | 1186 |
| Ndb_number_of_data_nodes | 4 |
+--------------------------+---------------+
If the MySQL server was built with clustering support, but it
is not connected to a cluster, all rows in the output of this
statement contain a zero or an empty string:
mysql> SHOW STATUS LIKE 'NDB%';
+--------------------------+-------+
| Variable_name | Value |
+--------------------------+-------+
| Ndb_cluster_node_id | 0 |
| Ndb_config_from_host | |
| Ndb_config_from_port | 0 |
| Ndb_number_of_data_nodes | 0 |
+--------------------------+-------+
See also Section 12.5.5.37, “SHOW STATUS Syntax”.
SELECT * FROM INFORMATION_SCHEMA.GLOBAL_STATUS WHERE
VARIABLE_NAME LIKE 'NDB%';
Beginning with MySQL 5.1.12, this statement provides similar
output to the SHOW
command discussed in the previous item. However, unlike the
case with SHOW STATUS , it is
possible using the SELECT to
extract values in SQL for use in scripts for monitoring and
automation purposes.
See Section 20.24, “The INFORMATION_SCHEMA GLOBAL_STATUS and
SESSION_STATUS
Tables”, for more information.
Beginning with MySQL Cluster NDB 7.1.1, you can also query the
tables in the ndbinfo information database for
real-time data about many MySQL Cluster operations. See
Section 17.5.8, “The ndbinfo MySQL Cluster Information Database”.
17.5.8. The ndbinfo MySQL Cluster Information Database
ndbinfo is a database storing containing
information specific to MySQL Cluster, available beginning with
MySQL Cluster NDB 7.1.1.
This database contains a number of tables, each providing a
different sort of data about MySQL Cluster node status, resource
usage, and operations. You can find more detailed information
about each of these tables in the next several sections.
ndbinfo is included with MySQL Cluster support
in the MySQL Server; no special compilation or configuration steps
are required; the tables are created by the MySQL Server when it
connects to the cluster. You can verify that
ndbinfo support is active in a given MySQL
Server instance by checking the output of
SHOW ENGINES for a line including
ndbinfo in the Engine column
and YES in the Support
column, as shown here (emphasized text):
mysql> SHOW ENGINES;
+------------+---------+----------------------------------------------------------------+--------------+------+------------+
| Engine | Support | Comment | Transactions | XA | Savepoints |
+------------+---------+----------------------------------------------------------------+--------------+------+------------+
| ndbcluster | YES | Clustered, fault-tolerant tables | YES | NO | NO |
| MRG_MYISAM | YES | Collection of identical MyISAM tables | NO | NO | NO |
| ndbinfo | YES | MySQL Cluster system information storage engine | NO | NO | NO |
| CSV | YES | CSV storage engine | NO | NO | NO |
| MEMORY | YES | Hash based, stored in memory, useful for temporary tables | NO | NO | NO |
| FEDERATED | NO | Federated MySQL storage engine | NULL | NULL | NULL |
| ARCHIVE | YES | Archive storage engine | NO | NO | NO |
| InnoDB | YES | Supports transactions, row-level locking, and foreign keys | YES | YES | YES |
| MyISAM | DEFAULT | Default engine as of MySQL 3.23 with great performance | NO | NO | NO |
| BLACKHOLE | YES | /dev/null storage engine (anything you write to it disappears) | NO | NO | NO |
+------------+---------+----------------------------------------------------------------+--------------+------+------------+
10 rows in set (0.00 sec)
You can also do this using SHOW
PLUGINS ; if ndbinfo support is
enabled, you should see a row containing
ndbinfo in the Name column
and ACTIVE in the Status
column, as shown here (emphasized text):
mysql> SHOW PLUGINS;
+------------+----------+----------------+---------+---------+
| Name | Status | Type | Library | License |
+------------+----------+----------------+---------+---------+
| binlog | ACTIVE | STORAGE ENGINE | NULL | GPL |
| partition | ACTIVE | STORAGE ENGINE | NULL | GPL |
| ARCHIVE | ACTIVE | STORAGE ENGINE | NULL | GPL |
| BLACKHOLE | ACTIVE | STORAGE ENGINE | NULL | GPL |
| CSV | ACTIVE | STORAGE ENGINE | NULL | GPL |
| FEDERATED | DISABLED | STORAGE ENGINE | NULL | GPL |
| MEMORY | ACTIVE | STORAGE ENGINE | NULL | GPL |
| InnoDB | ACTIVE | STORAGE ENGINE | NULL | GPL |
| MyISAM | ACTIVE | STORAGE ENGINE | NULL | GPL |
| MRG_MYISAM | ACTIVE | STORAGE ENGINE | NULL | GPL |
| ndbcluster | ACTIVE | STORAGE ENGINE | NULL | GPL |
| ndbinfo | ACTIVE | STORAGE ENGINE | NULL | GPL |
+------------+----------+----------------+---------+---------+
12 rows in set (0.00 sec)
If ndbinfo support is enabled, then you can
access ndbinfo via SQL statements in
mysql or another MySQL client. For example, you
can see ndbinfo listed in the output of
SHOW DATABASES , as shown here:
mysql> SHOW DATABASES;
+--------------------+
| Database |
+--------------------+
| information_schema |
| mysql |
| ndbinfo |
| test |
+--------------------+
4 rows in set (0.00 sec)
If the mysqld process was not started with the
--ndbcluster option,
ndbinfo is not available and is not displayed
by SHOW DATABASES . If
mysqld was formerly connected to a MySQL
Cluster but the cluster becomes unavailable (due to events such as
cluster shutdown, loss of network connectivity, and so forth),
ndbinfo and its tables remain visible, but an
attempt to access any tables (other than blocks
or config_params ) fails with Got
error 157 'Connection to NDB failed' from NDBINFO.
Note
With the exception of the blocks and
config_params tables, what we refer to as
ndbinfo “tables” are actually
views generated from internal NDB
tables not visible to the MySQL Server.
All ndbinfo tables are read-only.
You can select the ndbinfo database with a
USE statement, and then issue a
SHOW TABLES statement to obtain a
list of tables, just as for any other database, like this:
mysql> USE ndbinfo;
Database changed
mysql> SHOW TABLES;
+-------------------+
| Tables_in_ndbinfo |
+-------------------+
| blocks |
| config_params |
| counters |
| logbuffers |
| logspaces |
| memoryusage |
| nodes |
| pools |
| resources |
| transporters |
+-------------------+
10 rows in set (0.00 sec)
You can execute SELECT statements
against these tables, just as you would normally expect:
mysql> SELECT * FROM memoryusage;
+---------+--------------+------+-------+
| node_id | DATA_MEMORY | used | max |
+---------+--------------+------+-------+
| 1 | DATA_MEMORY | 3230 | 6408 |
| 2 | DATA_MEMORY | 3230 | 6408 |
| 1 | INDEX_MEMORY | 16 | 12832 |
| 2 | INDEX_MEMORY | 16 | 12832 |
+---------+--------------+------+-------+
4 rows in set (0.37 sec)
More complex queries are possible, as shown here:
mysql> SELECT SUM(used) as 'Data Memory Used, All Nodes'
> FROM memoryusage
> WHERE DATA_MEMORY = 'DATA_MEMORY';
+-----------------------------+
| Data Memory Used, All Nodes |
+-----------------------------+
| 6460 |
+-----------------------------+
1 row in set (0.37 sec)
mysql> SELECT SUM(max) as 'Total IndexMemory Available'
> FROM memoryusage
> WHERE DATA_MEMORY = 'INDEX_MEMORY';
+-----------------------------+
| Total IndexMemory Available |
+-----------------------------+
| 25664 |
+-----------------------------+
1 row in set (0.33 sec)
ndbinfo table and column names are case
sensitive (as is the name of the ndbinfo
database itself). These identifiers are in lower case. Trying to
use the wrong lettercase results in an error, as shown in this
example:
mysql> SELECT * FROM nodes;
+---------+--------+---------+-------------+
| node_id | uptime | status | start_phase |
+---------+--------+---------+-------------+
| 1 | 13602 | STARTED | 0 |
| 2 | 16 | STARTED | 0 |
+---------+--------+---------+-------------+
2 rows in set (0.04 sec)
mysql> SELECT * FROM Nodes;
ERROR 1146 (42S02): Table 'ndbinfo.Nodes' doesn't exist
17.5.8.1. The ndbinfo blocks Table
The blocks table is a static table which
simply contains the names and internal IDs of all NDB kernel
blocks (see NDB Kernel Blocks). It
is for use by the other ndbinfo tables (most
of which are actually views) in mapping block numbers to block
names for producing human-readable output.
The following table provides information about the columns in
the blocks table. For each column, the table
shows the name, data type, and a brief description. Additional
information can be found in the notes following the table.
Although this is a static table, its content could possibly vary
between different MySQL Cluster releases.
17.5.8.2. The ndbinfo config_params Table
The config_params table is a static table
which provides the names and internal ID numbers of all MySQL
Cluster configuration parameters.
The following table provides information about the columns in
the config_params table. For each column, the
table shows the name, data type, and a brief description.
Additional information can be found in the notes following the
table.
Although this is a static table, its content can vary between
MySQL Cluster installations, since supported parameters can vary
due to differences between software releases, cluster hardware
configurations, and other factors.
17.5.8.3. The ndbinfo counters Table
The counters table provides running totals of
events such as reads and writes for specific kernel blocks and
data nodes. Counts are kept from the most recent node start or
restart; a node start or restart resets all counters on that
node. Not all kernel blocks have all types of counters.
The following table provides information about the columns in
the counters table. For each column, the
table shows the name, data type, and a brief description.
Additional information can be found in the notes following the
table.
counter_name can be any one of the following:
ATTRINFO , TRANSACTIONS ,
COMMITS , READS ,
SIMPLE_READS , WRITES ,
ABORTS , TABLE_SCANS ,
RANGE_SCANS , or
OPERATIONS .
17.5.8.4. The ndbinfo logbuffers Table
The logbuffer table provides information on
MySQL Cluster log buffer usage.
The following table provides information about the columns in
the logbuffers table. For each column, the
table shows the name, data type, and a brief description.
17.5.8.5. The ndbinfo logspaces Table
This table provides information about MySQL Cluster log space
usage.
The following table provides information about the columns in
the logspaces table. For each column, the
table shows the name, data type, and a brief description.
17.5.8.6. The ndbinfo memoryusage Table
Querying this table provides information similar to that
provided by the ALL REPORT MemoryUsage
command in the ndb_mgm client, or logged by
ALL DUMP 1000 .
The following table provides information about the columns in
the memoryusage table. For each column, the
table shows the name, data type, and a brief description.
Additional information can be found in the notes following the
table.
For the used and max
columns, resources are measured in pages, which are 16K in size
for DataMemory and 8K for
IndexMemory . The next example shows how to
use this information in obtaining totals for
DataMemory , IndexMemory ,
or both. You can do this by summing across data nodes and table
columns, like this:
mysql> USE ndbinfo;
Database changed
mysql> SET @IPAGE := 8 * 1024;
Query OK, 0 rows affected (0.03 sec)
mysql> SET @DPAGE := 2 * @IPAGE;
Query OK, 0 rows affected (0.00 sec)
mysql> SELECT
-> @DPAGE * SUM(used) as 'Total DataMemory Used',
-> @DPAGE * SUM(max) as 'Total DataMemory Available',
-> @DPAGE * SUM(used) DIV COUNT(*) as 'Average DataMemory Used Per Node',
-> @DPAGE * SUM(max) DIV COUNT(*) as 'Average DataMemory Available Per Node'
-> FROM memoryusage WHERE memory_type = 'DATA_MEMORY'\G
*************************** 1. row ***************************
Total DataMemory Used: 106102784
Total DataMemory Available: 209977344
Average DataMemory Used Per Node: 53051392
Average DataMemory Available Per Node: 104988672
1 row in set (0.34 sec)
mysql> SELECT
-> @IPAGE * SUM(used) as 'Total IndexMemory Used',
-> @IPAGE * SUM(max) as 'Total IndexMemory Available',
-> @IPAGE * SUM(used) DIV COUNT(*) as 'Average IndexMemory Used Per Node',
-> @IPAGE * SUM(max) DIV COUNT(*) as 'Average IndexMemory Available Per Node'
-> FROM memoryusage WHERE memory_type = 'INDEX_MEMORY'\G
*************************** 1. row ***************************
Total IndexMemory Used: 376832
Total IndexMemory Available: 210239488
Average IndexMemory Used Per Node: 188416
Average IndexMemory Available Per Node: 105119744
1 row in set (0.33 sec)
(We use DIV rather than /
in both queries so that the results in all of the output columns
are integers.)
Prior to MySQL Cluster NDB 7.1.2, the
memory_type column was named
DATA_MEMORY . (Bug#50926)
17.5.8.7. The ndbinfo nodes Table
This table contains information on the status of data nodes. For
each data node that is running in the cluster, a corresponding
row in this table provides the node's node ID, status, and
uptime. For nodes that are starting, it also shows the current
start phase.
The following table provides information about the columns in
the nodes table. For each column, the table
shows the name, data type, and a brief description. Additional
information can be found in the notes following the table.
The uptime column shows the time in seconds
that this node has been running since it was last started or
restarted. This is a BIGINT
value. This figure includes the time actually needed to start
the node; in other words, this counter starts running the moment
that ndbd or ndbmtd is
first invoked; thus, even for a node that has not yet finished
starting, uptime may show a non-zero value.
The status column shows the node's
current status. This is one of: NOTHING ,
CMVMI , STARTING ,
STARTED , SINGLEUSER ,
STOPPING_1 , STOPPING_2 ,
STOPPING_3 , or STOPPING_4 .
When the status is STARTING , you can see the
current start phase in the start_phase column
(see later in this section). SINGLEUSER is
displayed in the status column for all data
nodes when the cluster is in single user mode (see
Section 17.5.6, “MySQL Cluster Single User Mode”). Seeing one of
the STOPPING states does not necessarily mean
that the node is shutting down but can mean rather that it is
entering a new state; for example, if you put the cluster in
single user mode, you can sometimes see data nodes report their
state briefly as STOPPING_2 before the status
changes to SINGLEUSER .
The start_phase column uses the same range of
values as those used in the output of the
ndb_mgm client
node_id STATUS
command (see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”). If the
node is not currently starting, then this column shows
0 . For a listing of MySQL Cluster start
phases with descriptions, see
Section 17.5.1, “Summary of MySQL Cluster Start Phases”.
Nodes that are stopped are not accounted for in this table.
Suppose that you have a MySQL Cluster with 2 data nodes (node
IDs 1 and 2), and all nodes are running normally, then this
table contains 2 rows, 1 for each data node:
mysql> USE ndbinfo;
Database changed
mysql> SELECT * FROM nodes;
+---------+--------+---------+-------------+
| node_id | uptime | status | start_phase |
+---------+--------+---------+-------------+
| 1 | 11776 | STARTED | 0 |
| 2 | 11761 | STARTED | 0 |
+---------+--------+---------+-------------+
2 rows in set (0.04 sec)
If you shut down one of the nodes, only the node that is still
running is reflected in the output of this
SELECT statement, as shown here:
ndb_mgm> 2 STOP
Node 2: Node shutdown initiated
Node 2: Node shutdown completed.
Node 2 has shutdown.
mysql> SELECT * FROM nodes;
+---------+--------+---------+-------------+
| node_id | uptime | status | start_phase |
+---------+--------+---------+-------------+
| 1 | 11807 | STARTED | 0 |
+---------+--------+---------+-------------+
1 row in set (0.02 sec)
17.5.8.8. The ndbinfo pools Table
This table provides information about
NDB memory pool usage.
The following table provides information about the columns in
the pools table. For each column, the table
shows the name, data type, and a brief description. Additional
information can be found in the notes following the table.
The columns param_name1 ,
param_name2 , param_name3 ,
and param_name4 contain names of
configuration parameters that are associated with a given pool.
If no parameters are associated with the pool, then all 4 of
these columns are NULL . For pools that are
associated with configuration parameters, these columns are not
filled with non-NULL values in any particular
order, as shown in this example:
mysql> SELECT
-> DISTINCT(pool_name), block_name, param_name1, param_name2, param_name3, param_name4
-> FROM ndbinfo.pools
-> WHERE pool_name = 'Table';
+-----------+------------+---------------+---------------+-----------------------+--------------------------+
| pool_name | block_name | param_name1 | param_name2 | param_name3 | param_name4 |
+-----------+------------+---------------+---------------+-----------------------+--------------------------+
| Table | BACKUP | NULL | MaxNoOfTables | MaxNoOfOrderedIndexes | MaxNoOfUniqueHashIndexes |
| Table | SUMA | MaxNoOfTables | NULL | NULL | NULL |
+-----------+------------+---------------+---------------+-----------------------+--------------------------+
2 rows in set (0.29 sec)
You can use a query such as the one shown here to generate a
table showing the associations of pools, NDB kernel blocks, and
(where applicable) configuration parameters:
SELECT
DISTINCT(pool_name) as 'Pool',
block_name as 'Block',
IF(
(param_name1 && param_name2 && param_name3 && param_name4) IS NULL,
'[NONE]',
REPLACE(
TRIM(
CONCAT(
IFNULL(param_name1, ''),
IFNULL(CONCAT(' ', param_name2), ''),
IFNULL(CONCAT(' ', param_name3), ''),
IFNULL(CONCAT(' ', param_name4), '')
)
), ' ', ', '
)
) as 'Parameter(s)'
FROM pools ORDER BY pool_name;
You may also find the following stored procedure useful. It
returns a list of pools associated with a given configuration
parameter, along with the kernel blocks to which these pools
belong. You can see here the statement used to create this
stored procedure along with sample output for some typical
parameters:
mysql> DELIMITER |
mysql> CREATE PROCEDURE get_pools (p VARCHAR(50))
-> BEGIN
-> SELECT
-> DISTINCT(pool_name) as Pool,
-> block_name as Block
-> FROM ndbinfo.pools
-> WHERE p
-> IN (param_name1, param_name2, param_name3, param_name4)
.> ORDER BY Pool;
-> END
-> |
Query OK, 0 rows affected (0.02 sec)
mysql> DELIMITER ;
mysql> CALL get_pools('MaxNoOfTables');
+-----------------+------------+
| pool_name | block_name |
+-----------------+------------+
| Table Record | DBDICT |
| DictObject | DBDICT |
| Index | DBTC |
| Table | BACKUP |
| Trigger | BACKUP |
| Fragment | BACKUP |
| Subscriber | SUMA |
| Table | SUMA |
| Subscription | SUMA |
| Index | DBTUX |
| Descriptor page | DBTUX |
+-----------------+------------+
11 rows in set (0.30 sec)
Query OK, 0 rows affected (0.30 sec)
mysql> CALL get_pools('MaxNoOfOrderedIndexes');
+-----------------+------------+
| pool_name | block_name |
+-----------------+------------+
| DictObject | DBDICT |
| Index | DBTC |
| Table | BACKUP |
| Trigger | BACKUP |
| Fragment | BACKUP |
| Index | DBTUX |
| Fragment | DBTUX |
| Descriptor page | DBTUX |
+-----------------+------------+
8 rows in set (0.29 sec)
17.5.8.9. The ndbinfo resources Table
This table provides information about data node resource
availability and usage.
These resources are sometimes known as
super-pools.
The following table provides information about the columns in
the resources table. For each column, the
table shows the name, data type, and a brief description.
Additional information can be found in the notes following the
table.
The resource_name can be one of
RESERVED , DISK_OPERATIONS ,
DISK_RECORDS , DATA_MEMORY ,
JOBBUFFER , FILE_BUFFERS ,
or TRANSPORTER_BUFFERS .
17.5.8.10. The ndbinfo transporters Table
This table contains information about NDB transporters.
The following table provides information about the columns in
the transporters table. For each column, the
table shows the name, data type, and a brief description.
Additional information can be found in the notes following the
table.
For each running data node in the cluster, the
transporters table displays a row showing the
status of each of that node's connections with all nodes in
the cluster, including itself. This
information is shown in the table's
status column, which can have any one of
the following values: CONNECTING ,
CONNECTED , DISCONNECTING ,
or DISCONNECTED .
Connections to API and management nodes which are configured but
not currently connected to the cluster are shown with status
DISCONNECTED . Rows where the
node_id is that of a data nodes which is not
currently connected are not shown in this table. (This is
similar omission of disconnected nodes in the
ndbinfo.nodes table — see
Section 17.5.8.7, “The ndbinfo nodes Table”, for more
information).
Assume you have a 5-node cluster conisting of 2 data nodes, 2
SQL nodes, and 1 management node, as shown in the output of the
SHOW command in the
ndb_mgm client:
ndb_mgm> SHOW
Connected to Management Server at: localhost:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=1 @10.100.10.1 (mysql-5.1.41 ndb-7.1.1, Nodegroup: 0, Master)
id=2 @10.100.10.2 (mysql-5.1.41 ndb-7.1.1, Nodegroup: 0)
[ndb_mgmd(MGM)] 1 node(s)
id=10 @10.100.10.10 (mysql-5.1.41 ndb-7.1.1)
[mysqld(API)] 2 node(s)
id=20 @10.100.10.20 (mysql-5.1.41 ndb-7.1.1)
id=21 @10.100.10.21 (mysql-5.1.41 ndb-7.1.1)
There are 10 rows in the transporters table
— 5 for the first data node, and 5 for the second —
assuming that all data nodes are running, as shown here:
mysql> SELECT * FROM transporters;
+---------+----------------+---------------+
| node_id | remote_node_id | status |
+---------+----------------+---------------+
| 1 | 1 | CONNECTED |
| 1 | 2 | CONNECTED |
| 1 | 10 | CONNECTED |
| 1 | 20 | CONNECTED |
| 1 | 21 | CONNECTED |
| 2 | 1 | CONNECTED |
| 2 | 2 | CONNECTED |
| 2 | 10 | CONNECTED |
| 2 | 20 | CONNECTED |
| 2 | 21 | CONNECTED |
+---------+----------------+---------------+
10 rows in set (0.04 sec)
If you shut down one of the data nodes in this cluster using the
ndb_mgm client STOP
command, then repeat the query, this table now shows only 5 rows
— 1 row for each connection from the remaining data node
to another node, including both itself and the data node that is
currently offline, as shown here:
mysql> SELECT * FROM transporters;
+---------+----------------+---------------+
| node_id | remote_node_id | status |
+---------+----------------+---------------+
| 1 | 1 | CONNECTED |
| 1 | 2 | CONNECTED |
| 1 | 10 | CONNECTED |
| 1 | 20 | CONNECTED |
| 1 | 21 | CONNECTED |
+---------+----------------+---------------+
5 rows in set (0.02 sec)
17.5.9. MySQL Cluster Security Issues
This section discusses security considerations to take into
account when setting up and running MySQL Cluster.
Topics to be covered in this chapter include the following:
MySQL Cluster and network security issues
Configuration issues relating to running MySQL Cluster
securely
MySQL Cluster and the MySQL privilege system
MySQL standard security procedures as applicable to MySQL
Cluster
17.5.9.1. MySQL Cluster Security and Networking Issues
In this section, we discuss basic network security issues as
they relate to MySQL Cluster. It is extremely important to
remember that MySQL Cluster “out of the box” is not
secure; you or your network administrator must take the proper
steps to ensure that your cluster cannot be compromised over the
network.
Cluster communication protocols are inherently insecure, and no
encryption or similar security measures are used in
communications between nodes in the cluster. Because network
speed and latency have a direct impact on the cluster's
efficiency, it is also not advisable to employ SSL or other
encryption to network connections between nodes, as such schemes
will effectively slow communications.
It is also true that no authentication is used for controlling
API node access to a MySQL Cluster. As with encryption, the
overhead of imposing authentication requirements would have an
adverse impact on Cluster performance.
In addition, there is no checking of the source IP address for
either of the following when accessing the cluster:
SQL or API nodes using “free slots” created
by empty [mysqld] or
[api] sections in the
config.ini file
This means that, if there are any empty
[mysqld] or [api]
sections in the config.ini file, then
any API nodes (including SQL nodes) that know the
management server's host name (or IP address) and port can
connect to the cluster and access its data without
restriction. (See
Section 17.5.9.2, “MySQL Cluster and MySQL Privileges”,
for more information about this and related issues.)
Note
You can exercise some control over SQL and API node
access to the cluster by specifying a
HostName parameter for all
[mysqld] and [api]
sections in the config.ini file.
However, this also means that, should you wish to
connect an API node to the cluster from a previously
unused host, you need to add an [api]
section containing its host name to the
config.ini file.
More information is available
elsewhere in this
chapter about the HostName
parameter. Also see
Section 17.3.1, “Quick Test Setup of MySQL Cluster”, for configuration
examples using HostName with API
nodes.
Any ndb_mgm client
This means that any cluster management client that is
given the management server's host name (or IP address)
and port (if not the standard port) can connect to the
cluster and execute any management client command. This
includes commands such as ALL STOP and
SHUTDOWN .
For these reasons, it is necessary to protect the cluster on the
network level. The safest network configuration for Cluster is
one which isolates connections between Cluster nodes from any
other network communications. This can be accomplished by any of
the following methods:
Keeping Cluster nodes on a network that is physically
separate from any public networks. This option is the most
dependable; however, it is the most expensive to
implement.
We show an example of a MySQL Cluster setup using such a
physically segregated network here:
This setup has two networks, one private (solid box) for
the Cluster management servers and data nodes, and one
public (dotted box) where the SQL nodes reside. (We show
the management and data nodes connected using a gigabit
switch since this provides the best performance.) Both
networks are protected from the outside by a hardware
firewall, sometimes also known as a
network-based firewall.
This network setup is safest because no packets can reach
the cluster's management or data nodes from outside the
network — and none of the cluster's internal
communications can reach the outside — without going
through the SQL nodes, as long as the SQL nodes do not
allow any packets to be forwarded. This means, of course,
that all SQL nodes must be secured against hacking
attempts.
Using one or more software firewalls (also known as
host-based firewalls) to control
which packets pass through to the cluster from portions of
the network that do not require access to it. In this type
of setup, a software firewall must be installed on every
host in the cluster which might otherwise be accessible
from outside the local network.
The host-based option is the least expensive to implement,
but relies purely on software to provide protection and so
is the most difficult to keep secure.
This type of network setup for MySQL Cluster is
illustrated here:
Using this type of network setup means that there are two
zones of MySQL Cluster hosts. Each cluster host must be
able to communicate with all of the other machines in the
cluster, but only those hosting SQL nodes (dotted box) can
be permitted to have any contact with the outside, while
those in the zone containing the data nodes and management
nodes (solid box) must be isolated from any machines that
are not part of the cluster. Applications using the
cluster and user of those applications must
not be permitted to have direct
access to the management and data node hosts.
To accomplish this, you must set up software firewalls
that limit the traffic to the type or types shown in the
following table, according to the type of node that is
running on each cluster host computer:
Any traffic other than that shown in the table for a given
node type should be denied.
The specifics of configuring a firewall vary from firewall
application to firewall application, and are beyond the
scope of this Manual. iptables is a
very common and reliable firewall application, which is
often used with APF as a front end to
make configuration easier. You can (and should) consult
the documentation for the software firewall that you
employ, should you choose to implement a MySQL Cluster
network setup of this type, or of a “mixed”
type as discussed under the next item.
It is also possible to employ a combination of the first
two methods, using both hardware and software to secure
the cluster — that is, using both network-based and
host-based firewalls. This is between the first two
schemes in terms of both security level and cost. This
type of network setup keeps the cluster behind the
hardware firewall, but allows incoming packets to travel
beyond the router connecting all cluster hosts in order to
reach the SQL nodes.
One possible network deployment of a MySQL Cluster using
hardware and software firewalls in combination is shown
here:
In this case, you can set the rules in the hardware
firewall to deny any external traffic except to SQL nodes
and API nodes, and then allow traffic to them only on the
ports required by your application.
Whatever network configuration you use, remember that your
objective from the viewpoint of keeping the cluster secure
remains the same — to prevent any unessential traffic from
reaching the cluster while ensuring the most efficient
communication between the nodes in the cluster.
Because MySQL Cluster requires large numbers of ports to be open
for communications between nodes, the recommended option is to
use a segregated network. This represents the simplest way to
prevent unwanted traffic from reaching the cluster.
Note
If you wish to administer a MySQL Cluster remotely (that is,
from outside the local network), the recommended way to do
this is to use ssh or another secure login
shell to access an SQL node host. From this host, you can then
run the management client to access the management server
safely, from within the Cluster's own local network.
Even though it is possible to do so in theory, it is
not recommended to use
ndb_mgm to manage a Cluster directly from
outside the local network on which the Cluster is running.
Since neither authentication nor encryption takes place
between the management client and the management server, this
represents an extremely insecure means of managing the
cluster, and is almost certain to be compromised sooner or
later.
17.5.9.2. MySQL Cluster and MySQL Privileges
In this section, we discuss how the MySQL privilege system works
in relation to MySQL Cluster and the implications of this for
keeping a MySQL Cluster secure.
Standard MySQL privileges apply to MySQL Cluster tables. This
includes all MySQL privilege types
(SELECT privilege,
UPDATE privilege,
DELETE privilege, and so on)
granted on the database, table, and column level. As with any
other MySQL Server, user and privilege information is stored in
the mysql system database. The SQL statements
used to grant and revoke privileges on
NDB tables, databases containing
such tables, and columns within such tables are identical in all
respects with the GRANT and
REVOKE statements used in
connection with database objects involving any (other) MySQL
storage engine. The same thing is true with respect to the
CREATE USER and
DROP USER statements.
It is important to keep in mind that the MySQL grant tables use
the MyISAM storage engine. Because of this,
those tables are not duplicated or shared among MySQL servers
acting as SQL nodes in a MySQL Cluster. By way of example,
suppose that two SQL nodes A
and B are connected to the same
MySQL Cluster, which has an NDB
table named mytable in a database named
mydb , and that you execute an SQL statement
on server A that creates a new
user jon@localhost and grants this user the
SELECT privilege on that table:
mysql> GRANT SELECT ON mydb.mytable
-> TO jon@localhost IDENTIFIED BY 'mypass';
This user is not created on server
B. In order for this to take
place, the statement must also be run on server
B. Similarly, statements run on
server A and affecting the
privileges of existing users on server
A do not affect users on server
B unless those statements are
actually run on server B as
well.
In other words, changes in users and their privileges
do not automatically propagate between SQL nodes.
Synchronization of privileges between SQL nodes must be done
either manually or by scripting an application that periodically
synchronizes the privilege tables on all SQL nodes in the
cluster.
Conversely, because there is no way in MySQL to deny privileges
(privileges can either be revoked or not granted in the first
place, but not denied as such), there is no special protection
for NDB tables on one SQL node from
users that have privileges on another SQL node. The most
far-reaching example of this is the MySQL
root account, which can perform any action on
any database object. In combination with empty
[mysqld] or [api] sections
of the config.ini file, this account can be
especially dangerous. To understand why, consider the following
scenario:
The config.ini file contains at least
one empty [mysqld] or
[api] section. This means that the
Cluster management server performs no checking of the host
from which a MySQL Server (or other API node) accesses the
MySQL Cluster.
There is no firewall, or the firewall fails to protect
against access to the Cluster from hosts external to the
network.
The host name or IP address of the Cluster's management
server is known or can be determined from outside the
network.
If these conditions are true, then anyone, anywhere can start a
MySQL Server with --ndbcluster
--ndb-connectstring=management_host
and access the Cluster. Using the MySQL root
account, this person can then perform the following actions:
Execute a SHOW DATABASES
statement to obtain a list of all databases that exist in
the cluster
Execute a SHOW TABLES FROM
some_database
statement to obtain a list of all
NDB tables in a given
database
Run any legal MySQL statements on any of those tables,
such as:
SELECT * FROM
some_table to
read all the data from any table
DELETE FROM
some_table to
delete all the data from a table
DESCRIBE
some_table or
SHOW CREATE TABLE
some_table to
determine the table schema
UPDATE
some_table SET
column1 =
any_value1 to
fill a table column with “garbage”
data; this could actually cause much greater damage
than simply deleting all the data
Even more insidious variations might include
statements like these:
UPDATE some_table SET an_int_column = an_int_column + 1
or
UPDATE some_table SET a_varchar_column = REVERSE(a_varchar_column )
Such malicious statements are limited only by the
imagination of the attacker.
The only tables that would be safe from this sort of
mayhem would be those tables that were created using
storage engines other than
NDB , and so not visible to a
“rogue” SQL node.
Note
A user who can log in as root can
also access the INFORMATION_SCHEMA
database and its tables, and so obtain information about
databases, tables, stored routines, scheduled events,
and any other database objects for which metadata is
stored in INFORMATION_SCHEMA .
It is also a very good idea to use different passwords for
the root accounts on different cluster
SQL nodes.
In sum, you cannot have a safe MySQL Cluster if it is directly
accessible from outside your local network.
Important
Never leave the MySQL root account password
empty. This is just as true when running MySQL as a
MySQL Cluster SQL node as it is when running it as a
standalone (non-Cluster) MySQL Server, and should be done as
part of the MySQL installation process before configuring the
MySQL Server as an SQL node in a MySQL Cluster.
You should never convert the system tables in the
mysql database to use the
NDB storage engine. There are a
number of reasons why you should not do this, but the most
important reason is this: Many of the SQL statements
that affect mysql tables storing information
about user privileges, stored routines, scheduled events, and
other database objects cease to function if these tables are
changed to use any storage engine other than
MyISAM . This is a consequence of
various MySQL Server internals which are not expected to change
in the foreseeable future.
If you need to synchronize mysql system
tables between SQL nodes, you can use standard MySQL replication
to do so, or employ a script to copy table entries between the
MySQL servers.
Summary.
The two most important points to remember regarding the MySQL
privilege system with regard to MySQL Cluster are:
Users and privileges established on one SQL node do not
automatically exist or take effect on other SQL nodes in
the cluster.
Conversely, removing a user or privilege on one SQL node
in the cluster does not remove the user or privilege
from any other SQL nodes.
Once a MySQL user is granted privileges on an
NDB table from one SQL node
in a MySQL Cluster, that user can “see” any
data in that table regardless of the SQL node from which
the data originated.
17.5.9.3. MySQL Cluster and MySQL Security Procedures
In this section, we discuss MySQL standard security procedures
as they apply to running MySQL Cluster.
In general, any standard procedure for running MySQL securely
also applies to running a MySQL Server as part of a MySQL
Cluster. First and foremost, you should always run a MySQL
Server as the mysql system user; this is no
different from running MySQL in a standard (non-Cluster)
environment. The mysql system account should
be uniquely and clearly defined. Fortunately, this is the
default behavior for a new MySQL installation. You can verify
that the mysqld process is running as the
system user mysql by using the system command
such as the one shown here:
shell> ps aux | grep mysql
root 10467 0.0 0.1 3616 1380 pts/3 S 11:53 0:00 \
/bin/sh ./mysqld_safe --ndbcluster --ndb-connectstring=localhost:1186
mysql 10512 0.2 2.5 58528 26636 pts/3 Sl 11:53 0:00 \
/usr/local/mysql/libexec/mysqld --basedir=/usr/local/mysql \
--datadir=/usr/local/mysql/var --user=mysql --ndbcluster \
--ndb-connectstring=localhost:1186 --pid-file=/usr/local/mysql/var/mothra.pid \
--log-error=/usr/local/mysql/var/mothra.err
jon 10579 0.0 0.0 2736 688 pts/0 S+ 11:54 0:00 grep mysql
If the mysqld process is running as any other
user than mysql , you should immediately shut
it down and restart it as the mysql user. If
this user does not exist on the system, the
mysql user account should be created, and
this user should be part of the mysql user
group; in this case, you should also make sure that the MySQL
DataDir on this system is owned by the
mysql user, and that the SQL node's
my.cnf file includes
user=mysql in the [mysqld]
section. Alternatively, you can start the server with
--user=mysql on the command line, but it is
preferable to use the my.cnf option, since
you might forget to use the command-line option and so have
mysqld running as another user
unintentionally. The mysqld_safe startup
script forces MySQL to run as the mysql user.
Important
Never run mysqld as the system root user.
Doing so means that potentially any file on the system can be
read by MySQL, and thus — should MySQL be compromised
— by an attacker.
As mentioned in the previous section (see
Section 17.5.9.2, “MySQL Cluster and MySQL Privileges”), you
should always set a root password for the MySQL Server as soon
as you have it running. You should also delete the anonymous
user account that is installed by default. You can accomplish
these tasks via the following statements:
shell> mysql -u root
mysql> UPDATE mysql.user
-> SET Password=PASSWORD('secure_password ')
-> WHERE User='root';
mysql> DELETE FROM mysql.user
-> WHERE User='';
mysql> FLUSH PRIVILEGES;
Be very careful when executing the
DELETE statement not to omit the
WHERE clause, or you risk deleting
all MySQL users. Be sure to run the
FLUSH
PRIVILEGES statement as soon as you have modified the
mysql.user table, so that the changes take
immediate effect. Without
FLUSH
PRIVILEGES , the changes do not take effect until the
next time that the server is restarted.
Note
Many of the MySQL Cluster utilities such as
ndb_show_tables,
ndb_desc, and
ndb_select_all also work without
authentication and can reveal table names, schemas, and data.
By default these are installed on Unix-style systems with the
permissions wxr-xr-x (755), which means
they can be executed by any user that can access the
mysql/bin directory.
See Section 17.4, “MySQL Cluster Programs”, for more
information about these utilities.
17.5.10. MySQL Cluster Disk Data Tables
Beginning with MySQL 5.1.6, it is possible to store the nonindexed
columns of NDB tables on disk, rather
than in RAM as with previous versions of MySQL Cluster.
As part of implementing MySQL Cluster Disk Data work, a number of
improvements were made in MySQL Cluster for the efficient handling
of very large amounts (terabytes) of data during node recovery and
restart. These include a “no-steal” algorithm for
synchronising a starting node with very large data sets. For more
information, see the paper
Recovery
Principles of MySQL Cluster 5.1, by MySQL
Cluster developers Mikael Ronstr?m and Jonas Oreland.
MySQL Cluster Disk Data performance can be influenced by a number
of configuration parameters. For information about these
parameters and their effects, see
MySQL
Cluster Disk Data configuration parameters and
MySQL
Cluster Disk Data storage and GCP Stop
errors
The performance of a MySQL Cluster that uses Disk Data storage can
also be greatly improved by separating data node file systems from
undo log files and tablespace data files, which can be done using
symbolic links. For more information, see
Section 17.5.10.2, “Using Symbolic Links with Disk Data Objects”.
17.5.10.1. MySQL Cluster Disk Data Objects
MySQL Cluster Disk Data storage is implemented using a number of
Disk Data objects. These include the
following:
Tablespaces act as containers for
other Disk Data objects.
Undo log files undo information
required for rolling back transactions.
One or more undo log files are assigned to a log
file group, which is then assigned to a
tablespace.
Data files store Disk Data table
data. A data file is assigned directly to a tablespace.
Undo log files and data files are actual files in the filesystem
of each data node; by default they are placed in
ndb_node_id _fs in
the DataDir specified in the MySQL
Cluster config.ini file, and where
node_id is the data node's node
ID. It is possible to place these elsewhere by specifying either
an absolute or relative path as part of the filename when
creating the undo log or data file. Statements that create these
files are shown later in this section.
MySQL Cluster tablespaces and log file groups are not
implemented as files.
Important
Although not all Disk Data objects are implemented as files,
they all share the same namespace. This means that
each Disk Data object must be uniquely
named (and not merely each Disk Data object of a given type).
For example, you cannot have a tablespace and a log file group
both named dd1 .
Assuming that you have already set up a MySQL Cluster with all
nodes (including management and SQL nodes) running MySQL 5.1.6
or newer, the basic steps for creating a Cluster table on disk
are as follows:
Create a log file group, and assign one or more undo log
files to it (an undo log file is also sometimes referred to
as an undofile).
Note
In MySQL 5.1 and later, undo log files are necessary only
for Disk Data tables. They are no longer used for
NDBCLUSTER tables that are
stored only in memory.
Create a tablespace; assign the log file group, as well as
one or more data files, to the tablespace.
Create a Disk Data table that uses this tablespace for data
storage.
Each of these tasks can be accomplished using SQL statements in
the mysql client or other MySQL client
application, as shown in the example that follows.
We create a log file group named lg_1
using CREATE LOGFILE GROUP . This log file
group is to be made up of two undo log files, which we name
undo_1.log and
undo_2.log , whose initial sizes are 16
MB and 12 MB, respectively. (The default initial size for an
undo log file is 128 MB.) Optionally, you can also specify a
size for the log file group's undo buffer, or allow it to
assume the default value of 8 MB. In this example, we set
the UNDO buffer's size at 2 MB. A log file group must be
created with an undo log file; so we add
undo_1.log to lg_1
in this CREATE LOGFILE GROUP statement:
CREATE LOGFILE GROUP lg_1
ADD UNDOFILE 'undo_1.log'
INITIAL_SIZE 16M
UNDO_BUFFER_SIZE 2M
ENGINE NDBCLUSTER;
To add undo_2.log to the log file
group, use the following ALTER LOGFILE
GROUP statement:
ALTER LOGFILE GROUP lg_1
ADD UNDOFILE 'undo_2.log'
INITIAL_SIZE 12M
ENGINE NDBCLUSTER;
Some items of note:
The .log file extension used here
is not required. We use it merely to make the log files
easily recognisable.
Every CREATE LOGFILE GROUP and
ALTER LOGFILE GROUP statement must
include an ENGINE clause. In MySQL
5.1 (including MySQL Cluster NDB 6.X and 7.X through
7.1), the permitted values for this clause are
NDBCLUSTER and
NDB .
Important
In MySQL 5.1.8 and later, there can exist only one log
file group in the same MySQL Cluster at any given
time.
When you add an undo log file to a log file group using
ADD UNDOFILE
'filename ' , a file
with the name filename is
created in the
ndb_node_id _fs
directory within the DataDir of each
data node in the cluster, where
node_id is the node ID of the
data node. Each undo log file is of the size specified
in the SQL statement. For example, if a MySQL Cluster
has 4 data nodes, then the ALTER LOGFILE
GROUP statement just shown creates 4 undo log
files, 1 each on in the data directory of each of the 4
data nodes; each of these files is named
undo_2.log and each file is 12 MB
in size.
UNDO_BUFFER_SIZE is limited by the
amount of system memory available.
For more information about the CREATE LOGFILE
GROUP statement, see
Section 12.1.14, “CREATE LOGFILE GROUP Syntax”. For more
information about ALTER LOGFILE
GROUP , see
Section 12.1.3, “ALTER LOGFILE GROUP Syntax”.
Now we can create a tablespace, which contains files to be
used by MySQL Cluster Disk Data tables for storing their
data. A tablespace is also associated with a particular log
file group. When creating a new tablespace, you must specify
the log file group which it is to use for undo logging; you
must also specify a data file. You can add more data files
to the tablespace after the tablespace is created; it is
also possible to drop data files from a tablespace (an
example of dropping data files is provided later in this
section).
Assume that we wish to create a tablespace named
ts_1 which uses lg_1
as its log file group. This tablespace is to contain two
data files named data_1.dat and
data_2.dat , whose initial sizes are 32
MB and 48 MB, respectively. (The default value for
INITIAL_SIZE is 128 MB.) We can do this
using two SQL statements, as shown here:
CREATE TABLESPACE ts_1
ADD DATAFILE 'data_1.dat'
USE LOGFILE GROUP lg_1
INITIAL_SIZE 32M
ENGINE NDBCLUSTER;
ALTER TABLESPACE ts_1
ADD DATAFILE 'data_2.dat'
INITIAL_SIZE 48M
ENGINE NDBCLUSTER;
The CREATE TABLESPACE statement creates a
tablespace ts_1 with the data file
data_1.dat , and associates
ts_1 with log file group
lg_1 . The ALTER
TABLESPACE adds the second data file
(data_2.dat ).
Some items of note:
As is the case with the .log file
extension used in this example for undo log files, there
is no special significance for the
.dat file extension; it is used
merely for easy recognition of data files.
When you add a data file to a tablespace using
ADD DATAFILE
'filename ' , a file
with the name filename is
created in the
ndb_node_id _fs
directory within the DataDir of each
data node in the cluster, where
node_id is the node ID of the
data node. Each undo log file is of the size specified
in the SQL statement. For example, if a MySQL Cluster
has 4 data nodes, then the ALTER
TABLESPACE statement just shown creates 4 undo
log files, 1 each on in the data directory of each of
the 4 data nodes; each of these files is named
data_2.dat and each file is 48 MB
in size.
All CREATE TABLESPACE and
ALTER TABLESPACE statements must
contain an ENGINE clause; only tables
using the same storage engine as the tablespace can be
created in the tablespace. In MySQL 5.1 (including MySQL
Cluster NDB 6.X and 7.X through 7.1), the only permitted
values for this clause are
NDBCLUSTER and
NDB .
For more information about the CREATE
TABLESPACE and ALTER
TABLESPACE statements, see
Section 12.1.18, “CREATE TABLESPACE Syntax”, and
Section 12.1.8, “ALTER TABLESPACE Syntax”.
Now it is possible to create a table whose nonindexed
columns are stored on disk in the tablespace
ts_1 :
CREATE TABLE dt_1 (
member_id INT UNSIGNED NOT NULL AUTO_INCREMENT PRIMARY KEY,
last_name VARCHAR(50) NOT NULL,
first_name VARCHAR(50) NOT NULL,
dob DATE NOT NULL,
joined DATE NOT NULL,
INDEX(last_name, first_name)
)
TABLESPACE ts_1 STORAGE DISK
ENGINE NDBCLUSTER;
The TABLESPACE ... STORAGE DISK option
tells the NDBCLUSTER storage
engine to use tablespace ts_1 for disk
data storage.
Note
Beginning with MySQL Cluster NDB 6.2.5 and MySQL Cluster
NDB 6.3.2, it is also possible to specify whether an
individual column is stored on disk or in memory by using
a STORAGE clause as part of the
column's definition in a CREATE
TABLE or ALTER
TABLE statement. STORAGE DISK
causes the column to be stored on disk, and
STORAGE MEMORY causes in-memory storage
to be used. See Section 12.1.17, “CREATE TABLE Syntax”, for more
information.
Once table ts_1 has been created as
shown, you can perform
INSERT ,
SELECT ,
UPDATE , and
DELETE statements on it just
as you would with any other MySQL table.
For table dt_1 as it has been defined
here, only the dob and
joined columns are stored on disk. This
is because there are indexes on the id ,
last_name , and
first_name columns, and so data belonging
to these columns is stored in RAM. In MySQL 5.1 (including
MySQL Cluster NDB 6.X and 7.X through 7.1), only nonindexed
columns can be held on disk; indexes and indexed column data
continue to be stored in memory. This tradeoff between the
use of indexes and conservation of RAM is something you must
keep in mind as you design Disk Data tables.
Performance note.
The performance of a cluster using Disk Data storage is
greatly improved if Disk Data files are kept on a separate
physical disk from the data node file system. This must be
done for each data node in the cluster to derive any
noticeable benefit.
You may use absolute and relative file system paths with
ADD UNDOFILE and ADD
DATAFILE . Relative paths are calculated relative to
the data node's data directory. You may also use symbolic links;
see Section 17.5.10.2, “Using Symbolic Links with Disk Data Objects”, for more
information and examples.
A log file group, a tablespace, and any Disk Data tables using
these must be created in a particular order. The same is true
for dropping any of these objects:
A log file group cannot be dropped as long as any
tablespaces are using it.
A tablespace cannot be dropped as long as it contains any
data files.
You cannot drop any data files from a tablespace as long as
there remain any tables which are using the tablespace.
Beginning with MySQL 5.1.12, it is no longer possible to
drop files created in association with a different
tablespace than the one with which the files were created.
(Bug#20053)
For example, to drop all the objects created so far in this
section, you would use the following statements:
mysql> DROP TABLE dt_1;
mysql> ALTER TABLESPACE ts_1
-> DROP DATAFILE 'data_2.dat'
-> ENGINE NDBCLUSTER;
mysql> ALTER TABLESPACE ts_1
-> DROP DATAFILE 'data_1.dat'
-> ENGINE NDBCLUSTER;
mysql> DROP TABLESPACE ts_1
-> ENGINE NDBCLUSTER;
mysql> DROP LOGFILE GROUP lg_1
-> ENGINE NDBCLUSTER;
These statements must be performed in the order shown, except
that the two ALTER TABLESPACE ... DROP
DATAFILE statements may be executed in either order.
You can obtain information about data files used by Disk Data
tables by querying the FILES table
in the INFORMATION_SCHEMA database. An extra
“NULL row” was added to this
table in MySQL 5.1.14 for providing additional information about
undo log files. For more information and examples of use, see
Section 20.21, “The INFORMATION_SCHEMA FILES Table”.
Beginning with MySQL Cluster NDB 6.3.27 and MySQL Cluster NDB
7.0.8, it is also possible to view information about allocated
and free disk space for each Disk Data table or table partition
using the ndb_desc utility. For more
information, see
Section 17.4.9, “ndb_desc — Describe NDB Tables”.
17.5.10.2. Using Symbolic Links with Disk Data Objects
The performance of a MySQL Cluster that uses Disk Data storage
can be greatly improved by separating data node file systems
from undo log files and tablespace data files and placing these
on different disks. Previous to MySQL Cluster NDB 6.2.17, MySQL
Cluster NDB 6.3.22, and MySQL Cluster NDB 6.4.3, there was no
direct support for this in MySQL Cluster, but it was possible to
achieve this separation using symbolic links as described in
this section.
Beginning with MySQL Cluster NDB 6.2.17, MySQL Cluster NDB
6.3.22, and MySQL Cluster NDB 6.4.3, the data node configuration
parameters FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles make the use of
symbolic links for this purpose unnecessary. For information
about these parameters, see
Disk Data filesystem parameters.
Each data node in the cluster creates a file system in the
directory named
ndb_node_id _fs
under the data node's
DataDir
as defined in the config.ini file. In this
example, we assume that each data node host has 3 disks, aliased
as /data0 , /data1 , and
/data2 , and that the cluster's
config.ini includes the following:
[ndbd default]
DataDir= /data0
Our objective is to place all Disk Data log files in
/data1 , and all Disk Data data files in
/data2 , on each data node host.
Note
In this example, we assume that the cluster's data node
hosts are all using Linux operating systems. For other
platforms, you may need to substitute you operating
system's commands for those shown here.
To accomplish this, perform the following steps:
Under the data node file system create symbolic links
pointing to the other drives:
shell> cd /data0/ndb_2_fs
shell> ls
D1 D10 D11 D2 D8 D9 LCP
shell> ln -s /data0 dnlogs
shell> ln -s /data1 dndata
You should now have two symbolic links:
shell> ls -l --hide=D*
lrwxrwxrwx 1 user group 30 2007-03-19 13:58 dndata -> /data1
lrwxrwxrwx 1 user group 30 2007-03-19 13:59 dnlogs -> /data2
We show this only for the data node with node ID 2;
however, you must do this for each
data node.
Now, in the mysql client, create a log
file group and tablespace using the symbolic links, as
shown here:
mysql> CREATE LOGFILE GROUP lg1
-> ADD UNDOFILE 'dnlogs/undo1.log'
-> INITIAL_SIZE 150M
-> UNDO_BUFFER_SIZE = 1M
-> ENGINE=NDBCLUSTER;
mysql> CREATE TABLESPACE ts1
-> ADD DATAFILE 'dndata/data1.log'
-> USE LOGFILE GROUP lg1
-> INITIAL_SIZE 1G
-> ENGINE=NDBCLUSTER;
Verify that the files were created and placed correctly as
shown here:
shell> cd /data1
shell> ls -l
total 2099304
-rw-rw-r-- 1 user group 157286400 2007-03-19 14:02 undo1.dat
shell> cd /data2
shell> ls -l
total 2099304
-rw-rw-r-- 1 user group 1073741824 2007-03-19 14:02 data1.dat
If you are running multiple data nodes on one host, you
must take care to avoid having them try to use the same
space for Disk Data files. You can make this easier by
creating a symbolic link in each data node filesystem.
Suppose you are using /data0 for both
data node filesystems, but you wish to have the Disk Data
files for both nodes on /data1 . In
this case, you can do something similar to what is shown
here:
shell> cd /data0
shell> ln -s ndb_2_fs/dd /data1/dn2
shell> ln -s ndb_3_fs/dd /data1/dn3
shell> ls -l --hide=D* ndb_2_fs
lrwxrwxrwx 1 user group 30 2007-03-19 14:22 dd -> /data1/dn2
shell> ls -l --hide=D* ndb_3_fs
lrwxrwxrwx 1 user group 30 2007-03-19 14:22 dd -> /data1/dn3
Now you can create a logfile group and tablespace using
the symbolic link, like this:
mysql> CREATE LOGFILE GROUP lg1
-> ADD UNDOFILE 'dd/undo1.log'
-> INITIAL_SIZE 150M
-> UNDO_BUFFER_SIZE = 1M
-> ENGINE=NDBCLUSTER;
mysql> CREATE TABLESPACE ts1
-> ADD DATAFILE 'dd/data1.log'
-> USE LOGFILE GROUP lg1
-> INITIAL_SIZE 1G
-> ENGINE=NDBCLUSTER;
Verify that the files were created and placed correctly as
shown here:
shell> cd /data1
shell> ls
dn2 dn3
shell> ls dn2
undo1.log data1.log
shell> ls dn3
undo1.log data1.log
17.5.10.3. MySQL Cluster Disk Data Storage Requirements
The following items apply to Disk Data storage requirements:
Variable-length columns of Disk Data tables take up a
fixed amount of space. For each row, this is equal to the
space required to store the largest possible value for
that column.
For general information about calculating these values,
see Section 10.5, “Data Type Storage Requirements”.
You can obtain an estimate the amount of space available
in data files and undo log files by querying the
INFORMATION_SCHEMA.FILES
table. For more information and examples, see
Section 20.21, “The INFORMATION_SCHEMA FILES Table”.
Note
The OPTIMIZE TABLE
statement does not have any effect on Disk Data tables.
In a Disk Data table, the first 256 bytes of a
TEXT or
BLOB column are stored in
memory; only the remainder is stored on disk.
Each row in a Disk Data table uses 8 bytes in memory to
point to the data stored on disk. This means that, in some
cases, converting an in-memory column to the disk-based
format can actually result in greater memory usage. For
example, convering a CHAR(4) column
from memory-based to disk-based format increases the
amount of DataMemory used per row from
4 to 8 bytes.
Important
Starting the cluster with the --initial
option does not remove Disk Data files.
You must remove these manually prior to performing an initial
restart of the cluster.
17.5.11. Adding MySQL Cluster Data Nodes Online
This section describes how to add MySQL Cluster data nodes
“online” — that is, without needing to shut
down the cluster completely and restart it as part of the process.
This capability is available in MySQL Cluster NDB 7.0 (beginning
with MySQL Cluster NDB 6.4.0) and later MySQL Cluster release
series.
Important
Currently, you must add new data nodes to a MySQL Cluster as
part of a new node group. In addition, it is not possible to
change the number of replicas (or the number of nodes per node
group) online.
17.5.11.1. Adding MySQL Cluster Data Nodes Online: General Issues
This section provides general information about the behavior of
and current limitations in adding MySQL Cluster nodes online.
Redistribution of Data.
The ability to add new nodes online includes a means to
reorganize NDBCLUSTER table data
and indexes so that they are distributed across all data
nodes, including the new ones. Table reorganization of both
in-memory and Disk Data tables is supported. This
redistribution does not currently include unique indexes (only
ordered indexes are redistributed) or BLOB
table data, but we are working to add redistibution of these
in the near future. The redistribution for
NDBCLUSTER tables already
existing before the new data nodes were added is not
automatic, but can be accomplished using simple SQL statements
in mysql or another MySQL client
application. However, all data and indexes added to tables
created after a new node group has been added are distributed
automatically among all cluster data nodes, including those
added as part of the new node group.
Partial starts.
It is possible to add a new node group without all of the new
data nodes being started. It is also possible to add a new
node group to a degraded cluster — that is, a cluster
that is only partially started, or where one or more data
nodes are not running. In the latter case, the cluster must
have enough nodes running to be viable before the new node
group can be added.
Effects on ongoing operations.
Normal DML operations using MySQL Cluster data are not
prevented by the creation or addition of a new node group, or
by table reorganization. However, it is not possible to
perform DDL concurrently with table reorganization —
that is, no other DDL statements can be issued while an
ALTER TABLE ...
REORGANIZE PARTITION statement is executing. In
addition, during the execution of ALTER TABLE ...
REORGANIZE PARTITION (or the execution of any other
DDL statement), it is not possible to restart cluster data
nodes.
Failure handling.
Failures of data nodes during node group creation and table
reorganization are handled as hown in the following table:
Dropping node groups.
The ndb_mgm client supports a DROP
NODEGROUP command, but it is possible to drop a node
group only when no data nodes in the node group contain any
data. Since there is currently no way to “empty”
a specific data node or node group, this command works only
the following two cases:
After issuing CREATE NODEGROUP in the
ndb_mgm client, but before issuing
any ALTER
ONLINE TABLE ... REORGANIZE PARTITION
statements in the mysql client.
After dropping all
NDBCLUSTER tables using
DROP TABLE .
TRUNCATE TABLE does not
work for this purpose because the data nodes continue to
store the table definitions.
17.5.11.2. Adding MySQL Cluster Data Nodes Online: Basic procedure
In this section, we list the basic steps required to add new
data nodes to a MySQL Cluster. For a detailed example, see
Section 17.5.11.3, “Adding MySQL Cluster Data Nodes Online: Detailed Example”.
Note
Beginning with MySQL Cluster NDB 7.0.4, this procedure applies
whether you are using ndbd or
ndbmtd binaries for the data node
processes. Previously, this did not work with multi-threaded
data nodes. (Bug#43108)
Assuming that you already have a running MySQL Cluster, adding
data nodes online requires the following steps:
Edit the cluster configuration
config.ini file, adding new
[ndbd] sections corresponding to the
nodes to be added. In the case where the cluster uses
multiple management servers, these changes need to be made
to all config.ini files used by the
management servers.
You must be careful that node IDs for any new data nodes
added in the config.ini file do not
overlap node IDs used by existing nodes. In the event that
you have API nodes using dynamically allocated node IDs and
these IDs match node IDs that you want to use for new data
nodes, it is possible to force any such API nodes to
“migrate”, as described later in this
procedure.
Perform a rolling restart of all MySQL Cluster management
servers.
Important
All management servers must be restarted with the
--reload or --initial
option to force the reading of the new configuration.
Perform a rolling restart of all existing MySQL Cluster data
nodes. It is not necessary (or usually even desirable) to
use --initial when restarting
the existing data nodes.
If you are using API nodes with dynamically allocated IDs
matching any node IDs that you wish to assign to new data
nodes, you must restart all API nodes (including SQL nodes)
before restarting any of the data nodes processes in this
step. This causes any API nodes with node IDs that were
previously not explicitly assigned to relinquish those node
IDs and acquire new ones.
Perform a rolling restart of any SQL or API nodes connected
to the MySQL Cluster.
Perform an initial start of the new data nodes.
Note
The new data nodes may be started in any order, and can
also be started concurrently, as long as they are started
after the rolling restarts of all existing nodes have been
completed and before proceeding to the next step.
Execute one or more CREATE NODEGROUP
commands in the MySQL Cluster management client to create
the new node group or node groups to which the new data
nodes will belong.
Redistribute the cluster's data among all data nodes
(including the new ones) by issuing an
ALTER ONLINE
TABLE ... REORGANIZE PARTITION statement in the
mysql client for each
NDBCLUSTER table.
Note
This needs to be done only for tables already existing at
the time the new node group is added. Data in tables
created after the new node group is added is distributed
automatically; however, data added to any given table
tbl that existed before the new nodes
were added is not distributed using the new nodes until
that table has been reorganized using
ALTER ONLINE
TABLE tbl REORGANIZE PARTITION .
Reclaim the space freed on the “old” nodes by
issuing, for each NDBCLUSTER
table, an OPTIMIZE TABLE
statement in the mysql client.
17.5.11.3. Adding MySQL Cluster Data Nodes Online: Detailed Example
In this section we provide a detailed example illustrating how
to add new MySQL Cluster data nodes online, starting with a
MySQL Cluster having 2 data nodes in a single node group and
concluding with a cluster having 4 data nodes in 2 node groups.
Starting configuration.
For purposes of illustration, we assume a minimal
configuration, and that the cluster uses a
config.ini file containing only the
following information:
[ndbd default]
DataMemory = 100M
IndexMemory = 100M
NoOfReplicas = 2
DataDir = /usr/local/mysql/var/mysql-cluster
[ndbd]
Id = 1
HostName = 192.168.0.1
[ndbd]
Id = 2
HostName = 192.168.0.2
[mgm]
HostName = 192.168.0.10
Id = 10
[api]
Id=20
HostName = 192.168.0.20
[api]
Id=21
HostName = 192.168.0.21
Note
We have left a gap in the sequence between data node IDs and
other nodes. This make it easier later to assign node IDs that
are not already in use to data nodes which are newly added.
We also assume that you have already started the cluster using
the appropriate command line or my.cnf
options, and that running
SHOW in the management
client produces output similar to what is shown here:
-- NDB Cluster -- Management Client --
ndb_mgm> SHOW
Connected to Management Server at: 192.168.0.10:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=1 @192.168.0.1 (5.1.41-ndb-7.0.12, Nodegroup: 0, Master)
id=2 @192.168.0.2 (5.1.41-ndb-7.0.12, Nodegroup: 0)
[ndb_mgmd(MGM)] 1 node(s)
id=10 @192.168.0.10 (5.1.41-ndb-7.0.12)
[mysqld(API)] 2 node(s)
id=20 @192.168.0.20 (5.1.41-ndb-7.0.12)
id=21 @192.168.0.21 (5.1.41-ndb-7.0.12)
Finally, we assume that the cluster contains a single
NDBCLUSTER table created as shown
here:
USE n;
CREATE TABLE ips (
id BIGINT NOT NULL AUTO_INCREMENT PRIMARY KEY,
country_code CHAR(2) NOT NULL,
type CHAR(4) NOT NULL,
ip_address varchar(15) NOT NULL,
addresses BIGINT UNSIGNED DEFAULT NULL,
date BIGINT UNSIGNED DEFAULT NULL
) ENGINE NDBCLUSTER;
The memory usage and related information shown later in this
section was generated after inserting approximately 50000 rows
into this table.
Note
In this example, we show the single-threaded
ndbd being used for the data node
processes. However — beginning with MySQL Cluster NDB
7.0.4 (Bug#43108) — you can also apply this example, if
you are using the multi-threaded ndbmtd by
substituting ndbmtd for
ndbd wherever it appears in the steps that
follow.
Step 1: Update configuration file.
Open the cluster global configuration file in a text editor
and add [ndbd] sections corresponding to
the 2 new data nodes. (We give these data nodes IDs 3 and 4,
and assume that they are to be run on host machines at
addresses 192.168.0.3 and 192.168.0.4, respectively.) After
you have added the new sections, the contents of the
config.ini file should look like what is
shown here, where the additions to the file are shown in bold
type:
[ndbd default]
DataMemory = 100M
IndexMemory = 100M
NoOfReplicas = 2
DataDir = /usr/local/mysql/var/mysql-cluster
[ndbd]
Id = 1
HostName = 192.168.0.1
[ndbd]
Id = 2
HostName = 192.168.0.2
[ndbd]
Id = 3
HostName = 192.168.0.3
[ndbd]
Id = 4
HostName = 192.168.0.4
[mgm]
HostName = 192.168.0.10
Id = 10
[api]
Id=20
HostName = 192.168.0.20
[api]
Id=21
HostName = 192.168.0.21
Once you have made the necessary changes, save the file.
Step 2: Restart the management server.
Restarting the cluster management server requires that you
issue separate commands to stop the management server and then
to start it again, as follows:
Stop the management server using the management client
STOP command, as shown here:
ndb_mgm> 10 STOP
Node 10 has shut down.
Disconnecting to allow Management Server to shutdown
shell>
Because shutting down the management server causes the
management client to terminate, you must start the
management server from the system shell. For simplicity, we
assume that config.ini is in the same
directory as the management server binary, but in practice,
you must supply the correct path to the configuration file.
You must also supply the --reload or
--initial option so that the management
server reads the new configuration from the file rather than
its configuration cache. If your shell's current
directory is also the same as the directory where the
management server binary is located, then you can invoke the
management server as shown here:
shell> ndb_mgmd -f config.ini --reload
2008-12-08 17:29:23 [MgmSrvr] INFO -- NDB Cluster Management Server. 5.1.41-ndb-7.0.12
2008-12-08 17:29:23 [MgmSrvr] INFO -- Reading cluster configuration from 'config.ini'
If you check the output of
SHOW in the management
client after restarting the ndb_mgm process,
you should now see something like this:
-- NDB Cluster -- Management Client --
ndb_mgm> SHOW
Connected to Management Server at: 192.168.0.10:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=1 @192.168.0.1 (5.1.41-ndb-7.0.12, Nodegroup: 0, Master)
id=2 @192.168.0.2 (5.1.41-ndb-7.0.12, Nodegroup: 0)
id=3 (not connected, accepting connect from 192.168.0.3)
id=4 (not connected, accepting connect from 192.168.0.4)
[ndb_mgmd(MGM)] 1 node(s)
id=10 @192.168.0.10 (5.1.41-ndb-7.0.12)
[mysqld(API)] 2 node(s)
id=20 @192.168.0.20 (5.1.41-ndb-7.0.12)
id=21 @192.168.0.21 (5.1.41-ndb-7.0.12)
Step 3: Perform a rolling restart of the existing data nodes.
This step can be accomplished entirely within the cluster
management client using the RESTART
command, as shown here:
ndb_mgm> 1 RESTART
Node 1: Node shutdown initiated
Node 1: Node shutdown completed, restarting, no start.
Node 1 is being restarted
ndb_mgm> Node 1: Start initiated (version 7.0.12)
Node 1: Started (version 7.0.12)
ndb_mgm> 2 RESTART
Node 2: Node shutdown initiated
Node 2: Node shutdown completed, restarting, no start.
Node 2 is being restarted
ndb_mgm> Node 2: Start initiated (version 7.0.12)
ndb_mgm> Node 2: Started (version 7.0.12)
Important
After issuing each X
RESTART command, wait until the management client
reports Node X : Started
(version ...) before proceeding
any further.
Step 4: Perform a rolling restart of all cluster API nodes.
Shut down and restart each MySQL server acting as an SQL node
in the cluster using mysqladmin shutdown
followed by mysqld_safe (or another startup
script). This should be similar to what is shown here, where
password is the MySQL
root password for a given MySQL server
instance:
shell> mysqladmin -uroot -ppassword shutdown
081208 20:19:56 mysqld_safe mysqld from pid file
/usr/local/mysql/var/tonfisk.pid ended
shell> mysqld_safe --ndbcluster --ndb-connectstring=192.168.0.10 &
081208 20:20:06 mysqld_safe Logging to '/usr/local/mysql/var/tonfisk.err'.
081208 20:20:06 mysqld_safe Starting mysqld daemon with databases
from /usr/local/mysql/var
Of course, the exact input and output depend on how and where
MySQL is installed on the system, as well as which options you
choose to start it (and whether or not some or all of these
options are specified in a my.cnf file).
Step 5: Perform an initial start of the new data nodes.
From a system shell on each of the hosts for the new data
nodes, start the data nodes as shown here, using the
--initial option:
shell> ndbd -c 192.168.0.10 --initial
Note
Unlike the case with restarting the existing data nodes, you
can start the new data nodes concurrently; you do not need to
wait for one to finish starting before starting the other.
Wait until both of the new data nodes have started
before proceeding with the next step. Once the new
data nodes have started, you can see in the output of the
management client SHOW
command that they do not yet belong to any node group (as
indicated with bold type here):
ndb_mgm> SHOW
Connected to Management Server at: 192.168.0.10:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=1 @192.168.0.1 (5.1.41-ndb-7.0.12, Nodegroup: 0, Master)
id=2 @192.168.0.2 (5.1.41-ndb-7.0.12, Nodegroup: 0)
id=3 @192.168.0.3 (5.1.41-ndb-7.0.12, no nodegroup)
id=4 @192.168.0.4 (5.1.41-ndb-7.0.12, no nodegroup)
[ndb_mgmd(MGM)] 1 node(s)
id=10 @192.168.0.10 (5.1.41-ndb-7.0.12)
[mysqld(API)] 2 node(s)
id=20 @192.168.0.20 (5.1.41-ndb-7.0.12)
id=21 @192.168.0.21 (5.1.41-ndb-7.0.12)
Step 6: Create a new node group.
You can do this by issuing a CREATE
NODEGROUP command in the cluster management client.
This command takes as its argument a comma-separated list of
the node IDs of the data nodes to be included in the new node
group, as shown here:
ndb_mgm> CREATE NODEGROUP 3,4
Nodegroup 1 created
By issuing SHOW again, you
can verify that data nodes 3 and 4 have joined the new node
group (again indicated in bold type):
ndb_mgm> SHOW
Connected to Management Server at: 192.168.0.10:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=1 @192.168.0.1 (5.1.41-ndb-7.0.12, Nodegroup: 0, Master)
id=2 @192.168.0.2 (5.1.41-ndb-7.0.12, Nodegroup: 0)
id=3 @192.168.0.3 (5.1.41-ndb-7.0.12, Nodegroup: 1)
id=4 @192.168.0.4 (5.1.41-ndb-7.0.12, Nodegroup: 1)
[ndb_mgmd(MGM)] 1 node(s)
id=10 @192.168.0.10 (5.1.41-ndb-7.0.12)
[mysqld(API)] 2 node(s)
id=20 @192.168.0.20 (5.1.41-ndb-7.0.12)
id=21 @192.168.0.21 (5.1.41-ndb-7.0.12)
Step 7: Redistribute cluster data.
When a node group is created, existing data and indexes are
not automatically distributed to the new node group's
data nodes, as you can see by issuing the appropriate
REPORT command in the management client:
ndb_mgm> ALL REPORT MEMORY
Node 1: Data usage is 5%(177 32K pages of total 3200)
Node 1: Index usage is 0%(108 8K pages of total 12832)
Node 2: Data usage is 5%(177 32K pages of total 3200)
Node 2: Index usage is 0%(108 8K pages of total 12832)
Node 3: Data usage is 0%(0 32K pages of total 3200)
Node 3: Index usage is 0%(0 8K pages of total 12832)
Node 4: Data usage is 0%(0 32K pages of total 3200)
Node 4: Index usage is 0%(0 8K pages of total 12832)
By using ndb_desc with the
-p option, which causes the output to include
partitioning information, you can see that the table still uses
only 2 partitions (in the Per partition info
section of the output, shown here in bold text):
shell> ndb_desc -c 192.168.0.10 -d n ips -p
-- ips --
Version: 1
Fragment type: 9
K Value: 6
Min load factor: 78
Max load factor: 80
Temporary table: no
Number of attributes: 6
Number of primary keys: 1
Length of frm data: 340
Row Checksum: 1
Row GCI: 1
SingleUserMode: 0
ForceVarPart: 1
FragmentCount: 2
TableStatus: Retrieved
-- Attributes --
id Bigint PRIMARY KEY DISTRIBUTION KEY AT=FIXED ST=MEMORY AUTO_INCR
country_code Char(2;latin1_swedish_ci) NOT NULL AT=FIXED ST=MEMORY
type Char(4;latin1_swedish_ci) NOT NULL AT=FIXED ST=MEMORY
ip_address Varchar(15;latin1_swedish_ci) NOT NULL AT=SHORT_VAR ST=MEMORY
addresses Bigunsigned NULL AT=FIXED ST=MEMORY
date Bigunsigned NULL AT=FIXED ST=MEMORY
-- Indexes --
PRIMARY KEY(id) - UniqueHashIndex
PRIMARY(id) - OrderedIndex
-- Per partition info --
Partition Row count Commit count Frag fixed memory Frag varsized memory
0 26086 26086 1572864 557056
1 26329 26329 1605632 557056
NDBT_ProgramExit: 0 - OK
You can cause the data to be redistributed among all of the data
nodes by performing, for each
NDBCLUSTER table, an ALTER
ONLINE TABLE ... REORGANIZE PARTITION statement in the
mysql client. After issuing the statement
ALTER ONLINE TABLE ips REORGANIZE PARTITION ,
you can see using ndb_desc that the data for
this table is now stored using 4 partitions, as shown here (with
the relevant portions of the output in bold type):
shell> ndb_desc -c 192.168.0.10 -d n ips -p
-- ips --
Version: 16777217
Fragment type: 9
K Value: 6
Min load factor: 78
Max load factor: 80
Temporary table: no
Number of attributes: 6
Number of primary keys: 1
Length of frm data: 341
Row Checksum: 1
Row GCI: 1
SingleUserMode: 0
ForceVarPart: 1
FragmentCount: 4
TableStatus: Retrieved
-- Attributes --
id Bigint PRIMARY KEY DISTRIBUTION KEY AT=FIXED ST=MEMORY AUTO_INCR
country_code Char(2;latin1_swedish_ci) NOT NULL AT=FIXED ST=MEMORY
type Char(4;latin1_swedish_ci) NOT NULL AT=FIXED ST=MEMORY
ip_address Varchar(15;latin1_swedish_ci) NOT NULL AT=SHORT_VAR ST=MEMORY
addresses Bigunsigned NULL AT=FIXED ST=MEMORY
date Bigunsigned NULL AT=FIXED ST=MEMORY
-- Indexes --
PRIMARY KEY(id) - UniqueHashIndex
PRIMARY(id) - OrderedIndex
-- Per partition info --
Partition Row count Commit count Frag fixed memory Frag varsized memory
0 12981 52296 1572864 557056
1 13236 52515 1605632 557056
2 13105 13105 819200 294912
3 13093 13093 819200 294912
NDBT_ProgramExit: 0 - OK
Note
Normally, ALTER
[ONLINE] TABLE table_name
REORGANIZE PARTITION is used with a list of
partition identifiers and a set of partition definitions to
create a new partitioning scheme for a table that has already
been explicitly partitioned. Its use here to redistribute data
onto a new MySQL Cluster node group is an exception in this
regard; when used in this way, only the name of the table is
used following the TABLE keyword, and no
other keywords or identifiers follow REORGANIZE
PARTITION .
Prior to MySQL Cluster NDB 6.4.3,
ALTER ONLINE
TABLE ... REORGANIZE PARTITION with no
partition_names INTO
(partition_definitions )
option did not work correctly with Disk Data tables or with
in-memory NDBCLUSTER tables
having one or more disk-based columns. (Bug#42549)
For more information, see Section 12.1.7, “ALTER TABLE Syntax”.
Also, for each table, the ALTER ONLINE TABLE
statement should be followed by an OPTIMIZE
TABLE to reclaim wasted space. You can obtain a list
of all NDBCLUSTER tables using the
following query against the
INFORMATION_SCHEMA.TABLES table:
SELECT TABLE_SCHEMA, TABLE_NAME
FROM INFORMATION_SCHEMA.TABLES
WHERE ENGINE = 'NDBCLUSTER';
Note
The INFORMATION_SCHEMA.TABLES.ENGINE value
for a MySQL Cluster table is always
NDBCLUSTER , regardless of whether
the CREATE TABLE statement used to create
the table (or ALTER TABLE statement used to
convert an existing table from a different storage engine)
used NDB or
NDBCLUSTER in its
ENGINE option.
You can see after performing these statements in the output of
ALL REPORT MEMORY that the data and indexes
are now redistributed between all cluster data nodes, as shown
here:
ndb_mgm> ALL REPORT MEMORY
Node 1: Data usage is 5%(176 32K pages of total 3200)
Node 1: Index usage is 0%(76 8K pages of total 12832)
Node 2: Data usage is 5%(176 32K pages of total 3200)
Node 2: Index usage is 0%(76 8K pages of total 12832)
Node 3: Data usage is 2%(80 32K pages of total 3200)
Node 3: Index usage is 0%(51 8K pages of total 12832)
Node 4: Data usage is 2%(80 32K pages of total 3200)
Node 4: Index usage is 0%(50 8K pages of total 12832)
It is not necessary to issue
ALTER ONLINE TABLE
... REORGANIZE PARTITION statements for
NDBCLUSTER tables created
after the new data nodes have been added;
data added to such tables is distributed among all data nodes
automatically. However, in
NDBCLUSTER tables that existed
prior to the addition of the new nodes,
neither existing nor new data is distributed using the new nodes
until these tables have been reorganized using
ALTER ONLINE TABLE
... REORGANIZE PARTITION .
Alternative procedure, without rolling restart.
It is possible to avoid the need for a rolling restart by
configuring the extra data nodes, but not starting them, when
first starting the cluster. This can be accomplished by using
the NodeGroup data node configuration
parameter in the config.ini file, as
shown here (note the section with bold text). We assume, as
before, that you wish to start with two data nodes —
nodes 1 and 2 — in one node group and later to expand
the cluster to four data nodes, by adding a second node group
consisting of nodes 3 and 4:
[ndbd default]
DataMemory = 100M
IndexMemory = 100M
NoOfReplicas = 2
DataDir = /usr/local/mysql/var/mysql-cluster
[ndbd]
Id = 1
HostName = 192.168.0.1
[ndbd]
Id = 2
HostName = 192.168.0.2
[ndbd]
Id = 3
HostName = 192.168.0.3
NodeGroup = 65535
[ndbd]
Id = 4
HostName = 192.168.0.4
NodeGroup = 65535
[mgm]
HostName = 192.168.0.10
Id = 10
[api]
Id=20
HostName = 192.168.0.20
[api]
Id=21
HostName = 192.168.0.21
In this case, you must perform the initial start of the cluster
using the --nowait-nodes option
with ndbd (or ndbmtd in
MySQL Cluster NDB 7.0.4 and later) for each of the data nodes
that you wish to have online immediately, so that the cluster
does not wait for the remaining nodes to start:
shell> ndbd -c 192.168.0.10 --initial --nowait-nodes=3,4
When you are ready to add the second node group, you need only
perform the following additional steps:
Start data nodes 3 and 4, invoking the data node process
once for each new node:
shell> ndbd -c 192.168.0.10 --initial
Issue the appropriate CREATE NODEGROUP
command in the management client:
ndb_mgm> CREATE NODEGROUP 3,4
In the mysql client, issue ALTER
ONLINE TABLE ... REORGANIZE PARTITION and
OPTIMIZE TABLE statements for each
existing NDBCLUSTER table. (As
noted elsewhere in this section, existing MySQL Cluster
tables cannot use the new nodes for data distribution until
this has been done.)
17.6. MySQL Cluster Replication
Previous to MySQL 5.1.6, asynchronous
replication, more usually referred to simply as
“replication”, was not available when using MySQL
Cluster. MySQL 5.1.6 introduces master-slave replication of this
type for MySQL Cluster databases. This section explains how to set
up and manage a configuration wherein one group of computers
operating as a MySQL Cluster replicates to a second computer or
group of computers. We assume some familiarity on the part of the
reader with standard MySQL replication as discussed elsewhere in
this Manual. (See Chapter 16, Replication).
Normal (non-clustered) replication involves a “master”
server and a “slave” server, the master being the
source of the operations and data to be replicated and the slave
being the recipient of these. In MySQL Cluster, replication is
conceptually very similar but can be more complex in practice, as it
may be extended to cover a number of different configurations
including replicating between two complete clusters. Although a
MySQL Cluster itself depends on the
NDBCLUSTER storage engine for
clustering functionality, it is not necessary to use the Cluster
storage engine on the slave. However, for maximum availability, it
is possible to replicate from one MySQL Cluster to another, and it
is this type of configuration that we discuss, as shown in the
following figure:
In this scenario, the replication process is one in which successive
states of a master cluster are logged and saved to a slave cluster.
This process is accomplished by a special thread known as the NDB
binlog injector thread, which runs on each MySQL server and produces
a binary log (binlog ). This thread ensures that
all changes in the cluster producing the binary log — and not
just those changes that are effected via the MySQL Server —
are inserted into the binary log with the correct serialization
order. We refer to the MySQL replication master and replication
slave servers as replication servers or replication nodes, and the
data flow or line of communication between them as a
replication channel.
For information about performing point-in-time recovery with MySQL
Cluster and MySQL Cluster Replication, see
Section 17.6.9.2, “Point-In-Time Recovery Using MySQL Cluster Replication”.
17.6.1. MySQL Cluster Replication — Abbreviations and Symbols
Throughout this section, we use the following abbreviations or
symbols for referring to the master and slave clusters, and to
processes and commands run on the clusters or cluster nodes:
17.6.2. MySQL Cluster Replication — Assumptions and General Requirements
A replication channel requires two MySQL servers acting as
replication servers (one each for the master and slave). For
example, this means that in the case of a replication setup with
two replication channels (to provide an extra channel for
redundancy), there will be a total of four replication nodes, two
per cluster.
Replication of a MySQL Cluster as described in this section and
those following is dependent on row-based replication. This means
that the replication master MySQL server must be started with
--binlog-format=ROW or
--binlog-format=MIXED , as described
in Section 17.6.6, “Starting MySQL Cluster Replication (Single Replication Channel)”. For
general information about row-based replication, see
Section 16.1.2, “Replication Formats”.
Important
If you attempt to use MySQL Cluster Replication with
--binlog-format=STATEMENT ,
replication fails to work properly because the
ndb_binlog_index table on the master and the
epoch column of the
ndb_apply_status table on the slave are not
updated (see
Section 17.6.4, “MySQL Cluster Replication Schema and Tables”). Instead,
only updates on the MySQL server acting as the replication
master propagate to the slave, and no updates from any other SQL
nodes on the master cluster are replicated.
In all MySQL Cluster NDB 6.x releases, the default value for the
--binlog-format option is
MIXED .
Each MySQL server used for replication in either cluster must be
uniquely identified among all the MySQL replication servers
participating in either cluster (you cannot have replication
servers on both the master and slave clusters sharing the same
ID). This can be done by starting each SQL node using the
--server-id=id
option, where id is a unique integer.
Although it is not strictly necessary, we will assume for purposes
of this discussion that all MySQL installations are the same
version.
In any event, both MySQL servers involved in replication must be
compatible with one another with respect to both the version of
the replication protocol used and the SQL feature sets which they
support; the simplest and easiest way to assure that this is the
case is to use the same MySQL version for all servers involved.
Note that in many cases it is not possible to replicate to a slave
running a version of MySQL with a lower version number than that
of the master — see
Section 16.3.2, “Replication Compatibility Between MySQL Versions”, for details.
We assume that the slave server or cluster is dedicated to
replication of the master, and that no other data is being stored
on it.
Note
It is possible to replicate a MySQL Cluster using
statement-based replication. However, in this case, the
following restrictions apply:
All updates to data rows on the cluster acting as the
master must be directed to a single MySQL server.
It is not possible to replicate a cluster using multiple
simultaneous MySQL replication processes.
Only changes made at the SQL level are replicated.
These are in addition to the other limitations of
statement-based replication as opposed to row-based replication;
see Section 16.1.2.1, “Comparison of Statement-Based and Row-Based Replication”, for more specific
information concerning the differences between the two
replication formats.
17.6.3. Known Issues in MySQL Cluster Replication
The following are known problems or issues when using replication
with MySQL Cluster in MySQL 5.1 (including MySQL Cluster NDB 6.X
and 7.X through 7.1):
Loss of master-slave connection.
A loss of connection can occur either between the
replication master SQL node and the replication slave SQL
node, or between the replication master SQL node and the
data nodes in the master cluster. In the latter case, this
can occur not only as a result of loss of physical
connection (for example, a broken network cable), but due to
the overflow of data node event buffers; if the SQL node is
too slow to respond, it may be dropped by the cluster (this
is controllable to some degree by adjusting the
MaxBufferedEpochs
and
TimeBetweenEpochs
configuration parameters). If this occurs, it is
entirely possible for new data to be inserted into the
master cluster without being recorded in the replication
master's binary log. For this reason, to
guarantee high availability, it is extremely important to
maintain a backup replication channel, to monitor the
primary channel, and to fail over to the secondary
replication channel when necessary to keep the slave cluster
synchronized with the master. MySQL Cluster is not designed
to perform such monitoring on its own; for this, an external
application is required.
Prior to MySQL 5.1.18, a MySQL Cluster replication slave
mysqld had no way of detecting that the
connection from the master had been interrupted. For this
reason, it was possible for the slave to become inconsistent
with the master.
Beginning with MySQL 5.1.18, the replication master issues a
“gap” event when connecting or reconnecting to
the master cluster. (A gap event is a type of “incident
event,” which indicates an incident that occurs that
affects the contents of the database but that cannot easily be
represented as a set of changes. Examples of incidents are
server crashes, database resynchronization, (some) software
updates, and (some) hardware changes.) When the slave
encounters a gap in the replication log, it stops with an
error message. This message is available in the output of
SHOW SLAVE STATUS , and
indicates that the SQL thread has stopped due to an incident
registered in the replication stream, and that manual
intervention is required. See
Section 17.6.8, “Implementing Failover with MySQL Cluster Replication”, for more
information about what to do in such circumstances.
Important
Because MySQL Cluster is not designed on its own to monitor
replication status or provide failover, if high availability
is a requirement for the slave server or cluster, then you
must set up multiple replication lines, monitor the master
mysqld on the primary replication line,
and be prepared fail over to a secondary line if and as
necessary. This must be done manually, or possibly by means
of a third-party application. For information about
implementing this type of setup, see
Section 17.6.7, “Using Two Replication Channels for MySQL Cluster Replication”,
and Section 17.6.8, “Implementing Failover with MySQL Cluster Replication”.
However, if you are replicating from a standalone MySQL
server to a MySQL Cluster, one channel is usually
sufficient.
Multi-byte character sets.
There are several known issues with regard to the use of
multi-byte characters sets with MySQL Cluster Replication.
See Bug#27404 (fixed in MySQL 5.1.21), Bug#29562, Bug#29563,
and Bug#29564 for more information.
Circular replication.
Prior to MySQL 5.1.18, circular replication was not
supported with MySQL Cluster replication, due to the fact
that all log events created in a particular MySQL Cluster
were wrongly tagged with the server ID of the MySQL server
used as master and not with the server ID of the originating
server.
Beginning with MySQL 5.1.18, this limitation is lifted, as
discussed in the next few paragraphs, in which we consider the
example of a replication setup involving three MySQL Clusters
numbered 1, 2, and 3, in which Cluster 1 acts as the
replication master for Cluster 2, Cluster 2 acts as the master
for Cluster 3, and Cluster 3 acts as the master for Cluster 1.
Each cluster has two SQL nodes, with SQL nodes A and B
belonging to Cluster 1, SQL nodes C and D belonging to Cluster
2, and SQL nodes E and F belonging to Cluster 3.
Circular replication using these clusters is supported as long
as:
This type of circular replication setup is shown in the
following diagram:
In this scenario, SQL node A in Cluster 1 replicates to SQL
node C in Cluster 2; SQL node C replicates to SQL node E in
Cluster 3; SQL node E replicates to SQL node A. In other
words, the replication line (indicated by the red arrows in
the diagram) directly connects all SQL nodes used as
replication masters and slaves.
It should also be possible to set up circular replication in
which not all master SQL nodes are also slaves, as shown here:
In this case, different SQL nodes in each cluster are used as
replication masters and slaves. However, you must
not start any of the SQL nodes using
--log-slave-updates (see the
description of
this option for more information). This type of
circular replication scheme for MySQL Cluster, in which the
line of replication (again indicated by the red arrows in the
diagram) is discontinuous, should be possible, but it should
be noted that it has not yet been thoroughly tested and must
therefore still be considered experimental.
Note
Beginning with MySQL 5.1.24, the
NDB storage engine uses
idempotent execution mode, which
suppresses duplicate-key and other errors that otherwise
break circular replication of MySQL Cluster. This is
equivalent to setting the global
slave_exec_mode system
variable to IDEMPOTENT . This is also
required for multi-master replication when using MySQL
Cluster. (Bug#31609)
It is not necessary to set
slave_exec_mode in MySQL
Cluster replication; MySQL Cluster does this automatically
for all NDB tables and ignores
any attempts to set this variable explicitly.
Cluster replication and primary keys.
In MySQL 5.1.6, only those NDB
tables having explicit primary keys could be replicated.
This limitation was lifted in MySQL 5.1.7. However, in the
event of a node failure, errors in replication of
NDB tables without primary keys
can still occur, due to the possibility of duplicate rows
being inserted in such cases. For this reason, it is highly
recommended that all NDB tables
being replicated have primary keys.
Restarting with --initial .
Restarting the cluster with the --initial
option causes the sequence of GCI and epoch numbers to start
over from 0 . (This is generally true of
MySQL Cluster and not limited to replication scenarios
involving Cluster.) The MySQL servers involved in
replication should in this case be restarted. After this,
you should use the RESET
MASTER and RESET
SLAVE statements to clear the invalid
ndb_binlog_index and
ndb_apply_status tables. respectively.
auto_increment_offset and
auto_increment_increment
variables.
The use of the
auto_increment_offset and
auto_increment_increment
server system variables is supported beginning with MySQL
5.1.20. Previously, these produced unpredictable results
when used with NDB tables or
MySQL Cluster replication.
Replication from NDBCLUSTER to other
storage engines.
If you attempt to replicate from a MySQL Cluster to a slave
that uses a storage engine that does not handle its own
binary logging, the replication process aborts with the
error Binary logging not possible ... Statement
cannot be written atomically since more than one engine
involved and at least one engine is self-logging
(Error 1595). It is possible to work
around this issue in one of the following ways:
Turn off binary logging on the slave.
This can be accomplished by setting
sql_log_bin = 0 .
Change the storage engine used for the
mysql.ndb_apply_status table.
Causing this table to use an engine that does not
handle its own binary logging can also eliminate the
conflict. This can be done by issuing a statement
such as ALTER TABLE mysql.ndb_apply_status
ENGINE=MyISAM on the slave. It is safe to
do this when using a
non-NDB storage engine
on the slave, since you do not then need to worry
about keeping multiple slave SQL nodes synchronized.
Filter out changes to the mysql.ndb_apply_status
table on the slave.
This can be done by starting the slave SQL node with
the option
--replicate-ignore-table=mysql.ndb_apply_status .
If you need for other tables to be ignored by
replication, you might wish to use an appropriate
--replicate-wild-ignore-table
option instead.
Important
You should not disable replication
or binary logging of
mysql.ndb_apply_status or change the
storage engine used for this table when replicating from
one MySQL Cluster to another. See Replication
and binary log filtering rules with replication between
MySQL Clusters elsewhere in this section for
details.
Replication from NDB to a different storage
engine.
For replication from NDB to a
different storage engine, the relationship between the two
databases must be a simple master-slave one. This means that
circular or master-master replication is not supported
between MySQL Cluster and other storage engines. However,
the MySQL Cluster database can simultaneously replicate to
multiple slave MySQL Cluster databases.
In addition, it is not possible to configure more than one
replication channel when replicating between MySQL Cluster and
a different storage engine. If MySQL Cluster is the master
then it is still possible to have more than one MySQL Server
maintain a binary log of all changes; however, for the slave
to change masters (fail over), the new master-slave
relationship must be explicitly defined on the slave.
Replication from NDB to a nontransactional storage
engine.
When replicating from NDB to a
nontransactional storage engine such as
MyISAM , you may encounter
unnecessary duplicate key errors when replicating
INSERT
... ON DUPLICATE KEY UPDATE statements. You can
suppress these in MySQL Cluster NDB 6.2 by using
--ndb-log-update-as-write=0 ,
which forces all columns from updated rows to be sent (and
not just those that were updated). For MySQL Cluster NDB
6.3.3 and later, there are additional way to determine
whether or not an update to the row on the master should be
applied on the slave mysqld; see
Section 17.6.11, “MySQL Cluster Replication Conflict Resolution”,
for more information about these methods.
You should also be aware that, when replicating from
NDB to a storage engine that does
not implement transactions (such as
MyISAM ), if the slave fails in
applying one or more rows changes from a transaction, it does
not roll back the rest of the transaction. Because of this, it
cannot be guaranteed that transactional consistency will be
maintained on the slave in this type of scenario.
Replication and binary log filtering rules with replication between
MySQL Clusters.
If you are using any of the options
--replicate-do-* ,
--replicate-ignore-* ,
--binlog-do-db , or
--binlog-ignore-db to filter
databases or tables being replicated, care must be taken not
to block replication or binary logging of the
mysql.ndb_apply_status , which is required
for replication between MySQL Clusters to operate properly.
In particular, you must keep in mind the following:
You should also remember that:
Note
If you are replicating a MySQL Cluster to a slave that
uses a storage engine other than
NDBCLUSTER , the
considerations just given previously may not apply. See
Replication from
NDBCLUSTER to other storage
engines elsewhere in this section for
details.
MySQL Cluster Replication and IPv6.
Currently, the NDB API and MGM API do not support IPv6.
However, beginning with MySQL Cluster NDB 6.4.1, MySQL
Servers — including those acting as SQL nodes in a
MySQL Cluster — can use IPv6 to contact other MySQL
Servers. This means that you can replicate between MySQL
Clusters using IPv6 to connect the master and slave SQL
nodes as shown by the dotted arrow in the following diagram:
However, all connections originating
within the MySQL Cluster — shown
in the diagram by solid arrows — must use IPv4.
All MySQL Cluster data nodes, management servers, and
management clients must be accessible from one another using
IPv4. In addition, SQL nodes must use IPv4 to communicate with
the cluster. There is not currently any support in the NDB and
MGM APIs for IPv6, which means that any applications written
using these APIs must also make all connections using IPv4.
17.6.4. MySQL Cluster Replication Schema and Tables
Replication in MySQL Cluster makes use of a number of dedicated
tables in the mysql database on each MySQL
Server instance acting as an SQL node in both the cluster being
replicated and the replication slave (whether the slave is a
single server or a cluster). These tables are created during the
MySQL installation process by the
mysql_install_db script, and include a table
for storing the binary log's indexing data. Since the
ndb_binlog_index table is local to each MySQL
server and does not participate in clustering, it uses the
MyISAM storage engine. This means that it must
be created separately on each mysqld
participating in the master cluster. (However, the binlog itself
contains updates from all MySQL servers in the cluster to be
replicated.) This table is defined as follows:
CREATE TABLE `ndb_binlog_index` (
`Position` BIGINT(20) UNSIGNED NOT NULL,
`File` VARCHAR(255) NOT NULL,
`epoch` BIGINT(20) UNSIGNED NOT NULL,
`inserts` BIGINT(20) UNSIGNED NOT NULL,
`updates` BIGINT(20) UNSIGNED NOT NULL,
`deletes` BIGINT(20) UNSIGNED NOT NULL,
`schemaops` BIGINT(20) UNSIGNED NOT NULL,
PRIMARY KEY (`epoch`)
) ENGINE=MYISAM DEFAULT CHARSET=latin1;
Important
Prior to MySQL 5.1.14, the ndb_binlog_index
table was known as binlog_index , and was kept
in a separate cluster database, which in
MySQL 5.1.7 and earlier was known as the
cluster_replication database. Similarly, the
ndb_apply_status and
ndb_schema tables were known as
apply_status and schema ,
and were also found in the cluster (earlier
cluster_replication ) database. However,
beginning with MySQL 5.1.14, all MySQL Cluster replication
tables reside in the mysql system database.
Information about how this change affects upgrades from MySQL
Cluster 5.1.13 and earlier to 5.1.14 and later versions can be
found in Section C.1.36, “Changes in MySQL 5.1.14 (05 December 2006)”.
Beginning with MySQL Cluster NDB 6.3.2, this table has been
changed to facilitate 3-way replication recovery. Two columns
orig_server_id and
orig_epoch have been added to this table; when
mysqld is started with the
--ndb-log-orig option, these columns store,
respectively, the ID of the server on which the event originated
and the epoch in which the event took place on the originating
server. In addition, the table's primary key now includes
these two columns. The modified table definition is shown here:
CREATE TABLE `ndb_binlog_index` (
`Position` BIGINT(20) UNSIGNED NOT NULL,
`File` VARCHAR(255) NOT NULL,
`epoch` BIGINT(20) UNSIGNED NOT NULL,
`inserts` INT(10) UNSIGNED NOT NULL,
`updates` INT(10) UNSIGNED NOT NULL,
`deletes` INT(10) UNSIGNED NOT NULL,
`schemaops` INT(10) UNSIGNED NOT NULL,
`orig_server_id` INT(10) UNSIGNED NOT NULL,
`orig_epoch` BIGINT(20) UNSIGNED NOT NULL,
`gci` INT(10) UNSIGNED NOT NULL,
PRIMARY KEY (`epoch`,`orig_server_id`,`orig_epoch`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1;
The gci column was added in MySQL Cluster NDB
6.2.6 and MySQL Cluster NDB 6.3.2.
The following figure shows the relationship of the MySQL Cluster
replication master server, its binlog injector thread, and the
mysql.ndb_binlog_index table.
An additional table, named ndb_apply_status , is
used to keep a record of the operations that have been replicated
from the master to the slave. Unlike the case with
ndb_binlog_index , the data in this table is not
specific to any one SQL node in the (slave) cluster, and so
ndb_apply_status can use the NDB
Cluster storage engine, as shown here:
CREATE TABLE `ndb_apply_status` (
`server_id` INT(10) UNSIGNED NOT NULL,
`epoch` BIGINT(20) UNSIGNED NOT NULL,
`log_name` VARCHAR(255) CHARACTER SET latin1 COLLATE latin1_bin NOT NULL,
`start_pos` BIGINT(20) UNSIGNED NOT NULL,
`end_pos` BIGINT(20) UNSIGNED NOT NULL,
PRIMARY KEY (`server_id`) USING HASH
) ENGINE=NDBCLUSTER DEFAULT CHARSET=latin1;
This table is populated only on slaves; on the master, no
DataMemory is allocated to it. However, the
table is populated from the master. For this
reason, this table must be replicated and any replication
filtering or binary log filtering rules that prevent this prevent
replication between clusters from operating properly. For more
information about potential problems arising from such filtering
rules, see Section 17.6.3, “Known Issues in MySQL Cluster Replication”.
The log_name , start_pos , and
end_pos columns were added in MySQL 5.1.18.
The ndb_binlog_index and
ndb_apply_status tables are created in the
mysql database because they should not be
replicated. No user intervention is normally required to create or
maintain either of them. Both the
ndb_binlog_index and the
ndb_apply_status tables are maintained by the
NDB injector thread. This keeps the
master mysqld process updated to changes
performed by the NDB storage engine.
The NDB binlog injector
thread receives events directly from the
NDB storage engine. The
NDB injector is responsible for
capturing all the data events within the cluster, and ensures that
all events which change, insert, or delete data are recorded in
the ndb_binlog_index table. The slave I/O
thread transfers the events from the master's binary log to the
slave's relay log.
However, it is advisable to check for the existence and integrity
of these tables as an initial step in preparing a MySQL Cluster
for replication. It is possible to view event data recorded in the
binary log by querying the
mysql.ndb_binlog_index table directly on the
master. This can be also be accomplished using the
SHOW BINLOG EVENTS statement on
either the replication master or slave MySQL servers. (See
Section 12.5.5.3, “SHOW BINLOG EVENTS Syntax”.)
You can also obtain useful information from the output of
SHOW ENGINE NDB
STATUS .
The ndb_schema table is used to track schema
changes made to NDB tables. It is
defined as shown here:
CREATE TABLE ndb_schema (
`db` VARBINARY(63) NOT NULL,
`name` VARBINARY(63) NOT NULL,
`slock` BINARY(32) NOT NULL,
`query` BLOB NOT NULL,
`node_id` INT UNSIGNED NOT NULL,
`epoch` BIGINT UNSIGNED NOT NULL,
`id` INT UNSIGNED NOT NULL,
`version` INT UNSIGNED NOT NULL,
`type` INT UNSIGNED NOT NULL,
PRIMARY KEY USING HASH (db,name)
) ENGINE=NDB DEFAULT CHARSET=latin1;
Unlike the two tables previously mentioned in this section, the
ndb_schema table is not visible either to MySQL
SHOW statements, or in any
INFORMATION_SCHEMA tables; however, it can be
seen in the output of ndb_show_tables, as shown
here:
shell> ndb_show_tables -t 2
id type state logging database schema name
4 UserTable Online Yes mysql def ndb_apply_status
5 UserTable Online Yes ndbworld def City
6 UserTable Online Yes ndbworld def Country
3 UserTable Online Yes mysql def NDB$BLOB_2_3
7 UserTable Online Yes ndbworld def CountryLanguage
2 UserTable Online Yes mysql def ndb_schema
NDBT_ProgramExit: 0 - OK
It is also possible to SELECT from
this table in mysql and other MySQL client
applications, as shown here:
mysql> SELECT * FROM mysql.ndb_schema WHERE name='City' \G
*************************** 1. row ***************************
db: ndbworld
name: City
slock:
query: alter table City engine=ndb
node_id: 4
epoch: 0
id: 0
version: 0
type: 7
1 row in set (0.00 sec)
This can sometimes be useful when debugging applications.
Note
When performing schema changes on
NDB tables, applications should
wait until the ALTER TABLE
statement has returned in the MySQL client connection that
issued the statement before attempting to use the updated
definition of the table.
The ndb_schema table was added in MySQL 5.1.8.
Beginning with MySQL 5.1.14, if either of the
ndb_apply_status or
ndb_schema tables does not exist on the slave,
it is created by ndb_restore. (Bug#14612)
Conflict resolution for MySQL Cluster Replication requires the
presence of an additional mysql.ndb_replication
table. Currently, this table must be created manually. For
details, see
Section 17.6.11, “MySQL Cluster Replication Conflict Resolution”.
17.6.5. Preparing the MySQL Cluster for Replication
Preparing the MySQL Cluster for replication consists of the
following steps:
Check all MySQL servers for version compatibility (see
Section 17.6.2, “MySQL Cluster Replication — Assumptions and General Requirements”).
Create a slave account on the master Cluster with the
appropriate privileges:
mysqlM > GRANT REPLICATION SLAVE
-> ON *.* TO 'slave_user '@'slave_host '
-> IDENTIFIED BY 'slave_password ';
In the previous statement,
slave_user is the slave account
user name, slave_host is the host
name or IP address of the replication slave, and
slave_password is the password to
assign to this account.
For example, to create a slave user account with the name
“myslave ,” logging in from the
host named “rep-slave ,” and
using the password “53cr37 ,”
use the following GRANT
statement:
mysqlM > GRANT REPLICATION SLAVE
-> ON *.* TO 'myslave'@'rep-slave'
-> IDENTIFIED BY '53cr37';
For security reasons, it is preferable to use a unique user
account — not employed for any other purpose — for
the replication slave account.
Configure the slave to use the master. Using the MySQL
Monitor, this can be accomplished with the
CHANGE MASTER TO statement:
mysqlS > CHANGE MASTER TO
-> MASTER_HOST='master_host ',
-> MASTER_PORT=master_port ,
-> MASTER_USER='slave_user ',
-> MASTER_PASSWORD='slave_password ';
In the previous statement,
master_host is the host name or IP
address of the replication master,
master_port is the port for the
slave to use for connecting to the master,
slave_user is the user name set up
for the slave on the master, and
slave_password is the password set
for that user account in the previous step.
For example, to tell the slave to replicate from the MySQL
server whose host name is
“rep-master ,” using the
replication slave account created in the previous step, use
the following statement:
mysqlS > CHANGE MASTER TO
-> MASTER_HOST='rep-master',
-> MASTER_PORT=3306,
-> MASTER_USER='myslave',
-> MASTER_PASSWORD='53cr37';
For a complete list of clauses that can be used with this
statement, see Section 12.6.2.1, “CHANGE MASTER TO Syntax”.
You can also configure the slave to use the master by setting
the corresponding startup options in the slave server's
my.cnf file. To configure the slave in
the same way as the preceding example
CHANGE MASTER TO statement, the
following information would need to be included in the slave's
my.cnf file:
[mysqld]
master-host=rep-master
master-port=3306
master-user=myslave
master-password=53cr37
For additional options that can be set in
my.cnf for replication slaves, see
Section 16.1.3, “Replication and Binary Logging Options and Variables”.
If the master cluster is already in use, you can create a
backup of the master and load this onto the slave to cut down
on the amount of time required for the slave to synchronize
itself with the master. If the slave is also running MySQL
Cluster, this can be accomplished using the backup and restore
procedure described in
Section 17.6.9, “MySQL Cluster Backups With MySQL Cluster Replication”.
ndb-connectstring=management_host [:port ]
In the event that you are not using MySQL
Cluster on the replication slave, you can create a backup with
this command on the replication master:
shellM > mysqldump --master-data=1
Then import the resulting data dump onto the slave by copying
the dump file over to the slave. After this, you can use the
mysql client to import the data from the
dumpfile into the slave database as shown here, where
dump_file is the name of the file
that was generated using mysqldump on the
master, and db_name is the name of
the database to be replicated:
shellS > mysql -u root -p db_name < dump_file
For a complete list of options to use with
mysqldump, see Section 4.5.4, “mysqldump — A Database Backup Program”.
Note
If you copy the data to the slave in this fashion, you
should make sure that the slave is started with the
--skip-slave-start option on the
command line, or else include
skip-slave-start in the slave's
my.cnf file to keep it from trying to
connect to the master to begin replicating before all the
data has been loaded. Once the data loading has completed,
follow the additional steps outlined in the next two
sections.
Ensure that each MySQL server acting as a replication master
is configured with a unique server ID, and with binary logging
enabled, using the row format. (See
Section 16.1.2, “Replication Formats”.) These options can be
set either in the master server's my.cnf
file, or on the command line when starting the master
mysqld process. See
Section 17.6.6, “Starting MySQL Cluster Replication (Single Replication Channel)”, for
information regarding the latter option.
17.6.6. Starting MySQL Cluster Replication (Single Replication Channel)
This section outlines the procedure for starting MySQL Cluster
replication using a single replication channel.
Start the MySQL replication master server by issuing this
command:
shellM > mysqld --ndbcluster --server-id=id \
--log-bin --binlog-format=ROW &
In the previous statement, id is
this server's unique ID (see
Section 17.6.2, “MySQL Cluster Replication — Assumptions and General Requirements”). This
starts the server's mysqld process with
binary logging enabled using the proper logging format.
Note
You can also start the master with
--binlog-format=MIXED , in
which case row-based replication is used automatically when
replicating between clusters.
Start the MySQL replication slave server as shown here:
shellS > mysqld --ndbcluster --server-id=id &
In the previous statement, id is
the slave server's unique ID. It is not necessary to enable
logging on the replication slave.
Note
You should use the --skip-slave-start
option with this command or else you should include
skip-slave-start in the slave server's
my.cnf file, unless you want
replication to begin immediately. With the use of this
option, the start of replication is delayed until the
appropriate START SLAVE
statement has been issued, as explained in Step 4 below.
It is necessary to synchronize the slave server with the
master server's replication binlog. If binary logging has not
previously been running on the master, run the following
statement on the slave:
mysqlS > CHANGE MASTER TO
-> MASTER_LOG_FILE='',
-> MASTER_LOG_POS=4;
This instructs the slave to begin reading the master's binary
log from the log's starting point. Otherwise — that is,
if you are loading data from the master using a backup —
see Section 17.6.8, “Implementing Failover with MySQL Cluster Replication”, for
information on how to obtain the correct values to use for
MASTER_LOG_FILE and
MASTER_LOG_POS in such cases.
Finally, you must instruct the slave to begin applying
replication by issuing this command from the
mysql client on the replication slave:
mysqlS > START SLAVE;
This also initiates the transmission of replication data from
the master to the slave.
It is also possible to use two replication channels, in a manner
simliar to the procedure described in the next section; the
differences between this and using a single replication channel
are covered in
Section 17.6.7, “Using Two Replication Channels for MySQL Cluster Replication”.
Beginning with MySQL Cluster NDB 6.2.3, it is possible to improve
cluster replication performance by enabling batched
updates. This can be accomplished by starting slave
mysqld processes with the
--slave-allow-batching option. Normally, updates
are applied as soon as they are received. However, the use of
batching causes updates to be applied in 32 KB batches, which can
result in higher throughput and less CPU usage, particularly where
individual updates are relatively small.
Note
Slave batching works on a per-epoch basis; updates belonging to
more than one transaction can be sent as part of the same batch.
All outstanding updates are applied when the end of an epoch is
reached, even if the updates total less than 32 KB.
Batching can be turned on and off at runtime. To activate it at
runtime, you can use either of these two statements:
SET GLOBAL slave_allow_batching = 1;
SET GLOBAL slave_allow_batching = ON;
If a particular batch causes problems (such as a statement whose
effects do not appear to be replicated correctly), slave batching
can be deactivated using either of the following statements:
SET GLOBAL slave_allow_batching = 0;
SET GLOBAL slave_allow_batching = OFF;
You can check whether slave batching is currently being used by
means of an appropriate SHOW
VARIABLES statement, like this one:
mysql> SHOW VARIABLES LIKE 'slave%';
+---------------------------+-------+
| Variable_name | Value |
+---------------------------+-------+
| slave_allow_batching | ON |
| slave_compressed_protocol | OFF |
| slave_load_tmpdir | /tmp |
| slave_net_timeout | 3600 |
| slave_skip_errors | OFF |
| slave_transaction_retries | 10 |
+---------------------------+-------+
6 rows in set (0.00 sec)
17.6.7. Using Two Replication Channels for MySQL Cluster Replication
In a more complete example scenario, we envision two replication
channels to provide redundancy and thereby guard against possible
failure of a single replication channel. This requires a total of
four replication servers, two masters for the master cluster and
two slave servers for the slave cluster. For purposes of the
discussion that follows, we assume that unique identifiers are
assigned as shown here:
Setting up replication with two channels is not radically
different from setting up a single replication channel. First, the
mysqld processes for the primary and secondary
replication masters must be started, followed by those for the
primary and secondary slaves. Then the replication processes may
be initiated by issuing the START
SLAVE statement on each of the slaves. The commands and
the order in which they need to be issued are shown here:
Start the primary replication master:
shellM > mysqld --ndbcluster --server-id=1 \
--log-bin --binlog-format=row &
Start the secondary replication master:
shellM' > mysqld --ndbcluster --server-id=2 \
--log-bin --binlog-format=row &
Start the primary replication slave server:
shellS > mysqld --ndbcluster --server-id=3 \
--skip-slave-start &
Start the secondary replication slave:
shellS' > mysqld --ndbcluster --server-id=4 \
--skip-slave-start &
Finally, initiate replication on the primary channel by
executing the START SLAVE
statement on the primary slave as shown here:
mysqlS > START SLAVE;
Warning
Only the primary channel is to be started at this point.
The secondary replication channel is to be started only in
the event that the primary replication channel fails, as
described in
Section 17.6.8, “Implementing Failover with MySQL Cluster Replication”.
Running multiple replication channels simultaneously can
result in unwanted duplicate records being created on the
replication slaves.
As mentioned previously, it is not necessary to enable binary
logging on replication slaves.
17.6.8. Implementing Failover with MySQL Cluster Replication
In the event that the primary Cluster replication process fails,
it is possible to switch over to the secondary replication
channel. The following procedure describes the steps required to
accomplish this.
Obtain the time of the most recent global checkpoint (GCP).
That is, you need to determine the most recent epoch from the
ndb_apply_status table on the slave
cluster, which can be found using the following query:
mysqlS' > SELECT @latest:=MAX(epoch)
-> FROM mysql.ndb_apply_status;
Using the information obtained from the query shown in Step 1,
obtain the corresponding records from the
ndb_binlog_index table on the master
cluster as shown here:
mysqlM' > SELECT
-> @file:=SUBSTRING_INDEX(File, '/', -1),
-> @pos:=Position
-> FROM mysql.ndb_binlog_index
-> WHERE epoch > @latest
-> ORDER BY epoch ASC LIMIT 1;
These are the records saved on the master since the failure of
the primary replication channel. We have employed a user
variable @latest here to represent the
value obtained in Step 1. Of course, it is not possible for
one mysqld instance to access user
variables set on another server instance directly. These
values must be “plugged in” to the second query
manually or in application code.
Now it is possible to synchronize the secondary channel by
running the following query on the secondary slave server:
mysqlS' > CHANGE MASTER TO
-> MASTER_LOG_FILE='@file',
-> MASTER_LOG_POS=@pos;
Again we have employed user variables (in this case
@file and @pos ) to
represent the values obtained in Step 2 and applied in Step 3;
in practice these values must be inserted manually or using
application code that can access both of the servers involved.
Note
@file is a string value such as
'/var/log/mysql/replication-master-bin.00001' ,
and so must be quoted when used in SQL or application code.
However, the value represented by @pos
must not be quoted. Although MySQL
normally attempts to convert strings to numbers, this case
is an exception.
You can now initiate replication on the secondary channel by
issuing the appropriate command on the secondary slave
mysqld:
mysqlS' > START SLAVE;
Once the secondary replication channel is active, you can
investigate the failure of the primary and effect repairs. The
precise actions required to do this will depend upon the reasons
for which the primary channel failed.
Warning
The secondary replication channel is to be started only if and
when the primary replication channel has failed. Running
multiple replication channels simultaneously can result in
unwanted duplicate records being created on the replication
slaves.
If the failure is limited to a single server, it should (in
theory) be possible to replicate from M
to S' , or from
M' to S ;
however, this has not yet been tested.
17.6.9. MySQL Cluster Backups With MySQL Cluster Replication
This section discusses making backups and restoring from them
using MySQL Cluster replication. We assume that the replication
servers have already been configured as covered previously (see
Section 17.6.5, “Preparing the MySQL Cluster for Replication”, and the
sections immediately following). This having been done, the
procedure for making a backup and then restoring from it is as
follows:
There are two different methods by which the backup may be
started.
Method A.
This method requires that the cluster backup process was
previously enabled on the master server, prior to
starting the replication process. This can be done by
including the following line in a
[mysql_cluster] section in the
my.cnf file , where
management_host is the IP
address or host name of the
NDB management server for
the master cluster, and port
is the management server's port number:
ndb-connectstring=management_host [:port ]
In this case, the backup can be started by executing
this statement on the replication master:
shellM > ndb_mgm -e "START BACKUP"
Method B.
If the my.cnf file does not specify
where to find the management host, you can start the
backup process by passing this information to the
NDB management client as
part of the START BACKUP command.
This can be done as shown here, where
management_host and
port are the host name and
port number of the management server:
shellM > ndb_mgm management_host :port -e "START BACKUP"
In our scenario as outlined earlier (see
Section 17.6.5, “Preparing the MySQL Cluster for Replication”),
this would be executed as follows:
shellM > ndb_mgm rep-master:1186 -e "START BACKUP"
Copy the cluster backup files to the slave that is being
brought on line. Each system running an
ndbd process for the master cluster will
have cluster backup files located on it, and
all of these files must be copied to the
slave to ensure a successful restore. The backup files can be
copied into any directory on the computer where the slave
management host resides, so long as the MySQL and NDB binaries
have read permissions in that directory. In this case, we will
assume that these files have been copied into the directory
/var/BACKUPS/BACKUP-1 .
It is not necessary that the slave cluster have the same
number of ndbd processes (data nodes) as
the master; however, it is highly recommended this number be
the same. It is necessary that the slave
be started with the --skip-slave-start
option, to prevent premature startup of the replication
process.
Create any databases on the slave cluster that are present on
the master cluster that are to be replicated to the slave.
Important
A CREATE DATABASE (or
CREATE
SCHEMA ) statement corresponding to each database
to be replicated must be executed on each SQL node in the
slave cluster.
Reset the slave cluster using this statement in the MySQL
Monitor:
mysqlS > RESET SLAVE;
It is important to make sure that the slave's
apply_status table does not contain any
records prior to running the restore process. You can
accomplish this by running this SQL statement on the slave:
mysqlS > DELETE FROM mysql.ndb_apply_status;
You can now start the cluster restoration process on the
replication slave using the ndb_restore
command for each backup file in turn. For the first of these,
it is necessary to include the -m option to
restore the cluster metadata:
shellS > ndb_restore -c slave_host :port -n node-id \
-b backup-id -m -r dir
dir is the path to the directory
where the backup files have been placed on the replication
slave. For the ndb_restore commands
corresponding to the remaining backup files, the
-m option should not be
used.
For restoring from a master cluster with four data nodes (as
shown in the figure in
Section 17.6, “MySQL Cluster Replication”) where the backup
files have been copied to the directory
/var/BACKUPS/BACKUP-1 , the proper
sequence of commands to be executed on the slave might look
like this:
shellS > ndb_restore -c rep-slave:1186 -n 2 -b 1 -m \
-r ./var/BACKUPS/BACKUP-1
shellS > ndb_restore -c rep-slave:1186 -n 3 -b 1 \
-r ./var/BACKUPS/BACKUP-1
shellS > ndb_restore -c rep-slave:1186 -n 4 -b 1 \
-r ./var/BACKUPS/BACKUP-1
shellS > ndb_restore -c rep-slave:1186 -n 5 -b 1 -e \
-r ./var/BACKUPS/BACKUP-1
Important
The -e (or
--restore-epoch ) option in the final
invocation of ndb_restore in this example
is required in order that the epoch is written to the slave
mysql.ndb_apply_status . Without this
information, the slave will not be able to synchronize
properly with the master. (See
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.)
Now you need to obtain the most recent epoch from the
ndb_apply_status table on the slave (as
discussed in
Section 17.6.8, “Implementing Failover with MySQL Cluster Replication”):
mysqlS > SELECT @latest:=MAX(epoch)
FROM mysql.ndb_apply_status;
Using @latest as the epoch value obtained
in the previous step, you can obtain the correct starting
position @pos in the correct binary log
file @file from the master's
mysql.ndb_binlog_index table using the
query shown here:
mysqlM > SELECT
-> @file:=SUBSTRING_INDEX(File, '/', -1),
-> @pos:=Position
-> FROM mysql.ndb_binlog_index
-> WHERE epoch > @latest
-> ORDER BY epoch ASC LIMIT 1;
In the event that there is currently no replication traffic,
you can get this information by running
SHOW MASTER STATUS on the
master and using the value in the Position
column for the file whose name has the suffix with the
greatest value for all files shown in the
File column. However, in this case, you
must determine this and supply it in the next step manually or
by parsing the output with a script.
Using the values obtained in the previous step, you can now
issue the appropriate CHANGE MASTER
TO statement in the slave's mysql
client:
mysqlS > CHANGE MASTER TO
-> MASTER_LOG_FILE='@file',
-> MASTER_LOG_POS=@pos;
Now that the slave “knows” from what point in
which binlog file to start reading data
from the master, you can cause the slave to begin replicating
with this standard MySQL statement:
mysqlS > START SLAVE;
To perform a backup and restore on a second replication channel,
it is necessary only to repeat these steps, substituting the host
names and IDs of the secondary master and slave for those of the
primary master and slave replication servers where appropriate,
and running the preceding statements on them.
For additional information on performing Cluster backups and
restoring Cluster from backups, see
Section 17.5.3, “Online Backup of MySQL Cluster”.
17.6.9.1. MySQL Cluster Replication — Automating Synchronization of the
Replication Slave to the Master Binary Log
It is possible to automate much of the process described in the
previous section (see
Section 17.6.9, “MySQL Cluster Backups With MySQL Cluster Replication”). The
following Perl script reset-slave.pl serves
as an example of how you can do this.
#!/user/bin/perl -w
# file: reset-slave.pl
# Copyright ©2005 MySQL AB
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to:
# Free Software Foundation, Inc.
# 59 Temple Place, Suite 330
# Boston, MA 02111-1307 USA
#
# Version 1.1
######################## Includes ###############################
use DBI;
######################## Globals ################################
my $m_host='';
my $m_port='';
my $m_user='';
my $m_pass='';
my $s_host='';
my $s_port='';
my $s_user='';
my $s_pass='';
my $dbhM='';
my $dbhS='';
####################### Sub Prototypes ##########################
sub CollectCommandPromptInfo;
sub ConnectToDatabases;
sub DisconnectFromDatabases;
sub GetSlaveEpoch;
sub GetMasterInfo;
sub UpdateSlave;
######################## Program Main ###########################
CollectCommandPromptInfo;
ConnectToDatabases;
GetSlaveEpoch;
GetMasterInfo;
UpdateSlave;
DisconnectFromDatabases;
################## Collect Command Prompt Info ##################
sub CollectCommandPromptInfo
{
### Check that user has supplied correct number of command line args
die "Usage:\n
reset-slave >master MySQL host< >master MySQL port< \n
>master user< >master pass< >slave MySQL host< \n
>slave MySQL port< >slave user< >slave pass< \n
All 8 arguments must be passed. Use BLANK for NULL passwords\n"
unless @ARGV == 8;
$m_host = $ARGV[0];
$m_port = $ARGV[1];
$m_user = $ARGV[2];
$m_pass = $ARGV[3];
$s_host = $ARGV[4];
$s_port = $ARGV[5];
$s_user = $ARGV[6];
$s_pass = $ARGV[7];
if ($m_pass eq "BLANK") { $m_pass = '';}
if ($s_pass eq "BLANK") { $s_pass = '';}
}
############### Make connections to both databases #############
sub ConnectToDatabases
{
### Connect to both master and slave cluster databases
### Connect to master
$dbhM
= DBI->connect(
"dbi:mysql:database=mysql;host=$m_host;port=$m_port",
"$m_user", "$m_pass")
or die "Can't connect to Master Cluster MySQL process!
Error: $DBI::errstr\n";
### Connect to slave
$dbhS
= DBI->connect(
"dbi:mysql:database=mysql;host=$s_host",
"$s_user", "$s_pass")
or die "Can't connect to Slave Cluster MySQL process!
Error: $DBI::errstr\n";
}
################ Disconnect from both databases ################
sub DisconnectFromDatabases
{
### Disconnect from master
$dbhM->disconnect
or warn " Disconnection failed: $DBI::errstr\n";
### Disconnect from slave
$dbhS->disconnect
or warn " Disconnection failed: $DBI::errstr\n";
}
###################### Find the last good GCI ##################
sub GetSlaveEpoch
{
$sth = $dbhS->prepare("SELECT MAX(epoch)
FROM mysql.ndb_apply_status;")
or die "Error while preparing to select epoch from slave: ",
$dbhS->errstr;
$sth->execute
or die "Selecting epoch from slave error: ", $sth->errstr;
$sth->bind_col (1, \$epoch);
$sth->fetch;
print "\tSlave Epoch = $epoch\n";
$sth->finish;
}
####### Find the position of the last GCI in the binlog ########
sub GetMasterInfo
{
$sth = $dbhM->prepare("SELECT
SUBSTRING_INDEX(File, '/', -1), Position
FROM mysql.ndb_binlog_index
WHERE epoch > $epoch
ORDER BY epoch ASC LIMIT 1;")
or die "Prepare to select from master error: ", $dbhM->errstr;
$sth->execute
or die "Selecting from master error: ", $sth->errstr;
$sth->bind_col (1, \$binlog);
$sth->bind_col (2, \$binpos);
$sth->fetch;
print "\tMaster bin log = $binlog\n";
print "\tMaster Bin Log position = $binpos\n";
$sth->finish;
}
########## Set the slave to process from that location #########
sub UpdateSlave
{
$sth = $dbhS->prepare("CHANGE MASTER TO
MASTER_LOG_FILE='$binlog',
MASTER_LOG_POS=$binpos;")
or die "Prepare to CHANGE MASTER error: ", $dbhS->errstr;
$sth->execute
or die "CHANGE MASTER on slave error: ", $sth->errstr;
$sth->finish;
print "\tSlave has been updated. You may now start the slave.\n";
}
# end reset-slave.pl
17.6.9.2. Point-In-Time Recovery Using MySQL Cluster Replication
Point-in-time recovery — that is,
recovery of data changes made since a given point in time
— is performed after restoring a full backup that returns
the server to its state when the backup was made. Performing
point-in-time recovery of MySQL Cluster tables with MySQL
Cluster and MySQL Cluster Replication can be accomplished using
a native NDB data backup (taken by
issuing CREATE BACKUP in the
ndb_mgm client) and restoring the
ndb_binlog_index table (from a dump made
using mysqldump).
To perform point-in-time recovery of MySQL Cluster, it is
necessary to follow the steps shown here:
Back up all NDB databases in the cluster,
using the START BACKUP command in the
ndb_mgm client (see
Section 17.5.3, “Online Backup of MySQL Cluster”).
At some later point, prior to restoring the cluster, make a
backup of the mysql.ndb_binlog_index
table. It is probably simplest to use
mysqldump for this task. Also back up the
binary log files at this time.
This backup should be updated regularly — perhaps even
hourly — depending on your needs.
(Catastrophic failure or error occurs.)
Locate the last known good backup.
Clear the data node filesystems (using
ndbd
--initial or, in MySQL Cluster
NDB 7.0 and later, ndbmtd
--initial ).
Note
MySQL Cluster Disk Data tablespace and log files are not
removed by --initial . You
must delete these manually.
Use DROP TABLE or
TRUNCATE TABLE with the
mysql.ndb_binlog_index table.
Execute ndb_restore, restoring all data.
You must include the
--restore_epoch option
when you run ndb_restore, so that the
ndb_apply_status table is populated
correctly. (See
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”, for
more information.)
Restore the ndb_binlog_index table from
the output of mysqldump and restore the
binary log files from backup, if necessary.
Find the epoch applied most recently — that is, the
maximum epoch column value in the
ndb_apply_status table — as the
user variable @LATEST_EPOCH (emphasized):
SELECT @LAST_EPOCH:=MAX(epoch)
FROM mysql.ndb_apply_status;
Find the latest binary log file
(@FIRST_FILE ) and position
(Position column value) within this file
that correspond to @LATEST_EPOCH in the
ndb_binlog_index table:
SELECT Position, @FIRST_FILE:=File
FROM mysql.ndb_binlog_index
WHERE epoch > @LAST_EPOCH ORDER BY epoch ASC LIMIT 1;
Using mysqlbinlog, replay the binary log
events from the given file and position up to the point of
the failure. (See Section 4.6.7, “mysqlbinlog — Utility for Processing Binary Log Files”.)
See also Section 6.4, “Point-in-Time (Incremental) Recovery Using the Binary Log”, for more
information about the binary log, replication, and incremental
recovery.
17.6.10. MySQL Cluster Replication — Multi-Master and Circular Replication
Beginning with MySQL 5.1.18, it is possible to use MySQL Cluster
in multi-master replication, including circular replication
between a number of MySQL Clusters.
Note
Prior to MySQL 5.1.18, multi-master replication including
circular replication was not supported with MySQL Cluster
replication. This was because log events created in a particular
MySQL Cluster were wrongly tagged with the server ID of the
master rather than the server ID of the originating server.
Circular replication example.
In the next few paragraphs we consider the example of a
replication setup involving three MySQL Clusters numbered 1, 2,
and 3, in which Cluster 1 acts as the replication master for
Cluster 2, Cluster 2 acts as the master for Cluster 3, and
Cluster 3 acts as the master for Cluster 1. Each cluster has two
SQL nodes, with SQL nodes A and B belonging to Cluster 1, SQL
nodes C and D belonging to Cluster 2, and SQL nodes E and F
belonging to Cluster 3.
Circular replication using these clusters is supported as long as:
This type of circular replication setup is shown in the following
diagram:
In this scenario, SQL node A in Cluster 1 replicates to SQL node C
in Cluster 2; SQL node C replicates to SQL node E in Cluster 3;
SQL node E replicates to SQL node A. In other words, the
replication line (indicated by the red arrows in the diagram)
directly connects all SQL nodes used as replication masters and
slaves.
It is also possible to set up circular replication in such a way
that not all master SQL nodes are also slaves, as shown here:
In this case, different SQL nodes in each cluster are used as
replication masters and slaves. However, you must
not start any of the SQL nodes using
--log-slave-updates (see the
description of
this option for more information). This type of circular
replication scheme for MySQL Cluster, in which the line of
replication (again indicated by the red arrows in the diagram) is
discontinuous, should be possible, but it should be noted that it
has not yet been thoroughly tested and must therefore still be
considered experimental.
Important
Beginning with MySQL 5.1.24, you should execute the following
statement before starting circular replication:
mysql> SET GLOBAL SLAVE_EXEC_MODE = 'IDEMPOTENT';
This is necessary to suppress duplicate-key and other errors
that otherwise break circular replication in MySQL Cluster.
IDEMPOTENT mode is also required for
multi-master replication when using MySQL Cluster. (Bug#31609)
See slave_exec_mode , for more
information.
Using NDB-native backup and restore to initialize a slave MySQL Cluster.
When setting up circular replication, it is possible to
initialize the slave cluster by using the management client
BACKUP command on one MySQL Cluster to create
a backup and then applying this backup on another MySQL Cluster
using ndb_restore. However, this does not
automatically create binary logs on the second MySQL
Cluster's SQL node acting as the replication slave. In
order to cause the binary logs to be created, you must issue a
SHOW TABLES statement on that SQL
node; this should be done prior to running
START SLAVE .
This is a known issue which we intend to address in a future
release.
Multi-master failover example.
In this section, we discuss failover in a multi-master MySQL
Cluster replication setup with three MySQL Clusters having
server IDs 1, 2, and 3. In this scenario, Cluster 1 replicates
to Clusters 2 and 3; Cluster 2 also replicates to Cluster 3.
This relationship is shown here:
In other words, data replicates from Cluster 1 to Cluster 3 via
2 different routes: directly, and by way of Cluster 2.
Not all MySQL servers taking part in multi-master replication must
act as both master and slave, and a given MySQL Cluster might use
different SQL nodes for diffferent replication channels. Such a
case is shown here:
MySQL servers acting as replication slaves must be run with the
--log-slave-updates option. Which
mysqld processes require this option is also
shown in the preceding diagram.
Note
Using the --log-slave-updates option has
no effect on servers not being run as replication slaves.
The need for failover arises when one of the replicating clusters
goes down. In this example, we consider the case where Cluster 1
is lost to service, and so Cluster 3 loses 2 sources of updates
from Cluster 1. Because replication between MySQL Clusters is
asynchronous, there is no guarantee that Cluster 3's updates
originating directly from Cluster 1 are more recent than those
received via Cluster 2. You can handle this by ensuring that
Cluster 3 catches up to Cluster 2 with regard to updates from
Cluster 1. In terms of MySQL servers, this means that you need to
replicate any outstanding updates from MySQL server C to server F.
On server C, perform the following queries:
mysqlC> SELECT @latest:=MAX(epoch)
-> FROM mysql.ndb_apply_status
-> WHERE server_id=1;
mysqlC> SELECT
-> @file:=SUBSTRING_INDEX(File, '/', -1),
-> @pos:=Position
-> FROM mysql.ndb_binlog_index
-> WHERE orig_epoch >= @latest
-> AND orig_server_id = 1
-> ORDER BY epoch ASC LIMIT 1;
Copy over the values for @file and
@pos manually from server C to server F
(or have your application perform the equivalent). Then, on server
F, execute the following CHANGE MASTER
TO statement:
mysqlF> CHANGE MASTER TO
-> MASTER_HOST = 'serverC'
-> MASTER_LOG_FILE='@file',
-> MASTER_LOG_POS=@pos;
Once this has been done, you can issue a
START SLAVE statement on MySQL
server F, and any missing updates originating from server B will
be replicated to server F.
17.6.11. MySQL Cluster Replication Conflict Resolution
When using a replication setup involving multiple masters
(including circular replication), it is possible that different
masters may try to update the same row on the slave with different
data. Conflict resolution in MySQL Cluster Replication provides a
means of resolving such conflicts by allowing a user defined
resolution column to be used to determine whether or not an update
to the row on a given master should be applied on the slave. (This
column is sometimes referred to as a “timestamp”
column, even though this column' type cannot be
TIMESTAMP , as explained later in
this section.) Different methods can be used to compare resolution
column values on the slave when conflicts occur, as explained
later in this section; the method used can be set on a per-table
basis.
Important
Conflict resolution as described in this section is always
applied on a row-by-row basis rather than a transactional basis.
In addition, it is the application's responsibility to
ensure that the resolution column is correctly populated with
relevant values, so that the resolution function can make the
appropriate choice when determining whether to apply an update.
Requirements.
Preparations for conflict resolution must be made on both the
master and the slave:
On the master writing the binlogs, you must determine
which columns are sent (all columns or only those that
have been updated). This is done for the MySQL Server as a
whole by applying the mysqld startup
option -–ndb-log-updated-only
(described later in this section) or on a per-table basis
by entries in the mysql.ndb_replication
table.
On the slave, you must determine which type of conflict
resolution to apply (“latest timestamp wins”,
“same timestamp wins”, or none). This is done
using the mysql.ndb_replication system
table, on a per-table basis.
If only some but not all columns are sent, then the master and
slave can diverge.
Note
We refer to the column used for determining updates as a
“timestamp” column, but the data type of this
column is never TIMESTAMP ;
rather, its data type should be
INT
(INTEGER ) or
BIGINT . This column should be
UNSIGNED and NOT NULL .
Master column control.
We can see update operations in terms of “before”
and “after” images — that is, the states of
the table before and after the update is applied. Normally, when
updating a table with a primary key, the “before”
image is not of great interest; however, when we need to
determine on a per-update basis whether or not to use the
updated values on a replication slave, we need to make sure that
both images are written to the master's binary log. This is
done with the
--ndb-log-update-as-write option
for mysqld, as described later in this
section.
Important
Whether logging of complete rows or of updated columns only
is done is decided when the MySQL server is started, and
cannot be changed online; you must either restart
mysqld, or start a new
mysqld instance with different logging
options.
Logging full or partial rows (--ndb-log-updated-only
option).
For purposes of conflict resolution, there are two basic methods
of logging rows, as determined by the setting of the
--ndb-log-updated-only option for
mysqld:
It is more efficient to log updated columns only; however, if
you need to log full rows, you can do so by setting
--ndb-log-updated-only to 0
or OFF .
Logging changed data as updates
(--ndb-log-update-as-write
option).
Either of these logging methods can be configured to be done
with or without the “before” image as determined by
the setting of another MySQL Server option
--ndb-log-update-as-write .
Because conflict resolution is done in the MySQL Server's update
handler, it is necessary to control logging on the master such
that updates are updates and not writes; that is, such that
updates are treated as changes in existing rows rather than the
writing of new rows (even though these replace existing rows).
This option is turned on by default; to turn it off, start the
server with --ndb-log-update-as-write=0 or
--ndb-log-update-as-write=OFF .
Conflict resolution control.
Conflict resolution is usually enabled on the server where
conflicts can occur. Like logging method selection, it is
enabled by entries in the
mysql.ndb_replication table.
The ndb_replication system table.
To enable conflict resolution, it is necessary to create an
ndb_replication table in the
mysql system database on the master, the
slave, or both, depending on the conflict resolution type and
method to be employed. This table is used to control logging and
conflict resolution functions on a per-table basis, and has one
row per table involved in replication.
ndb_replication is created and filled with
control information on the server where the conflict is to be
resolved. In a simple master-slave setup where data can also be
changed locally on the slave this will typically be the slave.
In a more complex master-master (2-way) replication schema this
will usually be all of the masters involved. Each row in
mysql.ndb_replication corresponds to a table
being replicated, and specifies how to log and resolve conflicts
(that is, which conflict resolution function, if any, to use)
for that table. The definition of the
mysql.ndb_replication table is shown here:
CREATE TABLE mysql.ndb_replication (
db VARBINARY(63),
table_name VARBINARY(63),
server_id INT UNSIGNED,
binlog_type INT UNSIGNED,
conflict_fn VARBINARY(128),
PRIMARY KEY USING HASH (db, table_name, server_id)
) ENGINE=NDB
PARTITION BY KEY(db,table_name);
The columns in this table are described in the following list:
db .
The name of the database containing the table to be
replicated.
table_name .
The name of the table to be replicated.
server_id .
The unique server ID of the MySQL instance (SQL node)
where the table resides.
binlog_type .
The type of binary logging to be employed. This is
determined as shown in the following table:
conflict_fn .
The conflict resolution function to be applied. This
function must be specified as one of the following:
NDB$OLD(column_name ).
If the value of
column_name is the
same on both the master and the slave, then the
update is applied; otherwise, the update is not
applied on the slave and an exception is written
to the log. This is illustrated by the following
pseudocode:
if (master_old_column_value == slave_current_column_value )
perform_update();
else
log_exception();
This function can be used for “same value
wins” conflict resolution. This type of
conflict resolution ensures that updates are not
applied on the slave from the wrong master.
Important
The column value from the master's
“before” image is used by this
function.
This conflict resolution function is available
beginning with MySQL Cluster NDB 6.3.4.
NDB$MAX(column_name ).
If the “timestamp” column value for
a given row coming from the master is higher
than that on the slave, it is applied; otherwise
it is not applied on the slave. This is
illustrated by the following pseudocode:
if (master_new_column_value > slave_current_column_value )
perform_update();
This function can be used for “greatest
timestamp wins” conflict resolution. This
type of conflict resolution ensures that, in the
event of a conflict, the version of the row that
was most recently updated is the version that
persists.
Important
The column value from the master's
“after” image is used by this
function.
This conflict resolution function is available
beginning with MySQL Cluster NDB 6.3.0.
NDB$MAX_DELETE_WIN(column_name ).
This is a variation on
NDB$MAX() . Due to the fact
that no timestamp is available for a delete
operation, a delete using
NDB$MAX() is in fact
processed as NDB$OLD . Howver,
for some use cases, this is not optimal. For
NDB$MAX_DELETE_WIN() , if the
“timestamp” column value for a
given row adding or updating an existing row
coming from the master is higher than that on
the slave, it is applied; however, delete
operations are treated as always having the
higher value. This is illustrated by the
following pseudocode:
if ( (master_new_column_value > slave_current_column_value )
||
operation.type == "delete")
perform_update();
This function can be used for “greatest
timestamp, delete wins” conflict
resolution. This type of conflict resolution
ensures that, in the event of a conflict, the
version of the row that was deleted or (otherwise)
most recently updated is the version that
persists.
This conflict resolution function is available
beginning with MySQL Cluster NDB 6.3.31 and MySQL
Cluster NDB 7.0.11.
Note
As with NDB$MAX() , the column
value from the master's
“after” image is the value used by
this function.
NULL .
Indicates that conflict resolution is not to be
used for the corresponding table.
.
Status information.
Beginning with MySQL Cluster NDB 6.3.3, a server status variable
Ndb_conflict_fn_max provides a
count of the number of times that a row was not applied on the
current SQL node due to “greatest timestamp wins”
conflict resolution since the last time that
mysqld was started.
Beginning with MySQL Cluster NDB 6.3.4, the number of times that a
row was not applied as the result of “same timestamp
wins” conflict resolution on a given
mysqld since the last time it was restarted is
given by the global status variable
Ndb_conflict_fn_old . In addition
to incrementing
Ndb_conflict_fn_old , the primary
key of the row that was not used is inserted into an
exceptions table, as explained later in
this section.
Additional requirements for “Same timestamp wins” conflict
resolution.
To use the NDB$OLD() conflict resolution
function, it is also necessary to create an exceptions table
corresponding to each NDB table for
which this type of conflict resolution is to be employed. The
name of this table is that of the table for which “same
timestamp wins” conflict resolution is to be applied,
with the string $EX appended. (For example,
if the name of the original table is mytable ,
the name of the corresponding exception table name should be
mytable$EX .) This table is created as
follows:
CREATE TABLE original_table $EX (
server_id INT UNSIGNED,
master_server_id INT UNSIGNED,
master_epoch BIGINT UNSIGNED,
count INT UNSIGNED,
original_table_pk_columns ,
[additional_columns ,]
PRIMARY KEY(server_id, master_server_id, master_epoch, count)
) ENGINE=NDB;
The first four columns are required. Following these columns,
the columns making up the original table's primary key should be
copied in the order in which they are used to define the primary
key of the original table.
Note
The names of the first four columns and the columns matching
the original table's primary key columns are not critical;
however, we suggest for reasons of clarity and consistency,
that you use the names shown here for the
server_id ,
master_server_id ,
master_epoch , and
count columns, and that you use the same
names as in the original table for the columns matching
those in the original table's primary key.
The data types for the columns duplicating the primary key
columns of the original table should be the same as for (or
larger than) the original columns.
Additional columns may optionally be defined following these
columns, but not before any of them; any such extra columns
cannot be NOT NULL . The exception
table's primary key must be defined as shown. The exception
table must use the NDB storage
engine. An example of use for NDB$OLD() and
an exception table is given later in this section.
Important
The mysql.ndb_replication table is read when
a data table is set up for replication, so the row corresponding
to a table to be replicated must be inserted into
mysql.ndb_replication
before the table to be replicated is
created.
Examples.
The following examples assume that you have already a working
MySQL Cluster replication setup, as described in
Section 17.6.5, “Preparing the MySQL Cluster for Replication”, and
Section 17.6.6, “Starting MySQL Cluster Replication (Single Replication Channel)”.
NDB$MAX() example.
Suppose you wish to enable “greatest timestamp
wins” conflict resolution on table
test.t1 , using column
mycol as the
“timestamp”. This can be done using the
following steps:
Make sure that you have started the master
mysqld with
-–ndb-log-update-as-write=OFF .
On the master, perform this
INSERT statement:
INSERT INTO mysql.ndb_replication
VALUES ('test', 't1', 0, NULL, 'NDB$MAX(mycol)');
Inserting a 0 into the
server_id
indicates that all SQL nodes accessing this table
should use conflict resolution. If you want to use
conflict resolution on a specific
mysqld only, use the actual
server ID.
Inserting NULL into the
binlog_type column has the same
effect as inserting 0
(NBT_DEFAULT ); the server
default is used.
Create the test.t1 table:
CREATE TABLE test.t1 (
columns
mycol INT UNSIGNED,
columns
) ENGINE=NDB;
Now, when updates are done on this table, conflict
resolution will be applied, and the version of the
row having the greatest value for
mycol will be written to the
slave.
Note
Other binlog_type options —
such as NBT_UPDATED_ONLY_USE_UPDATE
should be used in order to control logging on the master
via the ndb_replication table rather
than by using command-line options.
NDB$OLD() example.
Suppose an NDB table such
as the one defined here is being replicated, and you
wish to enable “same timestamp wins”
conflict resolution for updates to this table:
CREATE TABLE test.t2 (
a INT UNSIGNED NOT NULL,
b CHAR(25) NOT NULL,
columns ,
mycol INT UNSIGNED NOT NULL,
columns ,
PRIMARY KEY pk (a, b)
) ENGINE=NDB;
The following steps are required, in the order shown:
First — and prior to
creating test.t2 — you
must insert a row into the
mysql.ndb_replication table, as
shown here:
INSERT INTO mysql.ndb_replication
VALUES ('test', 't2', 0, NULL, 'NDB$OLD(mycol)');
Possible values for the
binlog_type column are shown
earlier in this section. The value
'NDB$OLD(mycol)' should be
inserted into the conflict_fn
column.
Create an appropriate exceptions table for
test.t2 . The table creation
statement shown here includes all required
columns; any additional columns must be declared
following these columns, and before the definition
of the table's primary key.
CREATE TABLE test.t2$EX (
server_id SMALLINT UNSIGNED,
master_server_id INT UNSIGNED,
master_epoch BIGINT UNSIGNED,
count BIGINT UNSIGNED,
a INT UNSIGNED NOT NULL,
b CHAR(25) NOT NULL,
[additional_columns ,]
PRIMARY KEY(server_id, master_server_id, master_epoch, count)
) ENGINE=NDB;
Create the table test.t2 as
shown previously.
These steps must be followed for every table for which
you wish to perform conflict resolution using
NDB$OLD() . For each such table, there
must be a corresponding row in
mysql.ndb_replication , and there must
be an exceptions table in the same database as the table
being replicated.
17.7. Changes in MySQL Cluster NDB 6.X and 7.X
This section contains changelog information for MySQL Cluster
releases that use versions 6.1, 6.2, 6.3, and 7.0 of the
NDBCLUSTER storage engine.
Each MySQL Cluster release is based on a mainline MySQL 5.1 release
and a particular version of the
NDBCLUSTER storage engine, as shown in
the version string returned by executing SELECT
VERSION() in the mysql client, or by
executing the ndb_mgm client
SHOW or
STATUS command; for more information, see
Chapter 17, MySQL Cluster NDB 6.X/7.X.
For general information about features added in MySQL Cluster, see
Section 17.1.4, “MySQL Cluster Development History”. For a complete list of
all bugfixes and feature changes in MySQL Cluster, please refer to
the changelog section for each individual MySQL Cluster release.
An overview of features added in MySQL 5.1 not specific to MySQL
Cluster can be found here: Section 1.5, “What Is New in MySQL 5.1”. For a
complete list of all bugfixes and features changes made in MySQL 5.1
that are not specific to MySQL Cluster, see
Section C.1, “Changes in Release 5.1.x (Production)”.
This section contains changelogs for individual MySQL Cluster NDB
6.X and 7.X releases.
For long-format changelogs covering each MySQL Cluster NDB 6.X and
7.X series, see Section 17.7.6, “Release Series Changelogs — MySQL Cluster NDB 6.X and 7.X”.
17.7.1. Changes in MySQL Cluster NDB 7.1
This section contains change history information for MySQL Cluster
releases based on version 7.1 of the
NDBCLUSTER storage engine, currently
in development.
For an overview of new features added in MySQL Cluster NDB 7.1,
see Section 17.1.4.6, “MySQL Cluster Development in MySQL Cluster NDB 7.1”.
17.7.1.1. Changes in MySQL Cluster NDB 7.1.2 (5.1.41-ndb-7.1.2) (Not yet released)
MySQL Cluster NDB 7.1.2 is a new beta release of MySQL Cluster,
incorporating new features in the
NDBCLUSTER storage engine and
fixing recently discovered bugs in MySQL Cluster NDB 7.1.1 and
previous MySQL Cluster releases.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.41 (see
Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Functionality added or changed: Bugs fixed: Important Change:
The DATA_MEMORY column of the
ndbinfo.memoryusage table was renamed to
memory_type .
(Bug#50926)
When using NoOfReplicas equal to 1 or 2, if
data nodes from one node group were restarted 256 times and
applications were running traffic such that it would encounter
NDB error 1204
(Temporary failure, distribution
changed), the live node in the node group would
crash, causing the cluster to crash as well. The crash occurred
only when the error was encountered on the 256th restart; having
the error on any previous or subsequent restart did not cause
any problems.
(Bug#50930)
ndbmtd started on a single-core machine could
sometimes fail with a Job Buffer Full
error when MaxNoOfExecutionThreads was set
greater than LockExecuteThreadToCPU . Now a
warning is logged when this occurs.
(Bug#50582)
17.7.1.2. Changes in MySQL Cluster NDB 7.1.1 (5.1.41-ndb-7.1.1) (01 February 2010 beta)
This is the first public beta release for MySQL Cluster NDB 7.1.
MySQL Cluster NDB 7.1.1 is a new beta release of MySQL Cluster,
incorporating new features in the
NDBCLUSTER storage engine and
fixing recently discovered bugs in MySQL Cluster NDB 7.0 and
previous MySQL Cluster releases.
Obtaining MySQL Cluster NDB 7.1.1.
MySQL Cluster NDB 7.1.1 is a source-only release, which can be
downloaded from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/mysql-5.1.41-ndb-7.1.1-beta.tgz.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.41 (see
Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Functionality added or changed:
The ndbinfo database is added to provide
MySQL Cluster metadata in real time. The tables making up this
database contain information about memory, buffer, and other
resource usage, as well as configuration paramaters and
settings, event counts, and other useful data. Access to
ndbinfo is done by executing standard SQL
queries on its tables using the mysql
command-line client or other MySQL client application. No
special setup procedures are required;
ndbinfo is created automatically and visible
in the output of SHOW DATABASES
when the MySQL Server is connected to a MySQL Cluster.
For more information about ndbinfo , see
Section 17.5.8, “The ndbinfo MySQL Cluster Information Database”.
Cluster API:
ClusterJ 1.0 and ClusterJPA 1.0 are now available for
programming Java applications with MySQL Cluster. ClusterJ is a
Java connector providing an object-relational API for performing
high-speed operations such as primary key lookups on a MySQL
Cluster database, but does not require the use of the MySQL
Server or JDBC (Connector/J). ClusterJ uses a new library
NdbJTie which allows direct access from Java to the NDB API and
thus to the NDBCLUSTER storage
engine. ClusterJPA is a new implementation of
OpenJPA, and can
use either a JDBC connection to a MySQL Cluster SQL node (MySQL
Server) or a direct connection to MySQL Cluster via NdbJTie,
depending on availability and operational performance.
ClusterJ, ClusterJPA, and NdbJTie require Java 1.5 or 1.6, and
MySQL Cluster NDB 7.0 or later.
All necessary libraries and other files for ClusterJ,
ClusterJPA, and NdbJTie can be found in the MySQL Cluster NDB
7.1.1 or later distribution.
17.7.1.3. Changes in MySQL Cluster NDB 7.1.0 (5.1.39-ndb-7.1.0) (Not released alpha)
This version was for testing and internal use only, and not
officially released.
Functionality added or changed: Important Change:
The default value of the DiskIOThreadPool
data node configuration parameter has changed from 8 to 2.
The NDBCLUSTER storage engine now
provides native support for default values of table columns.
This means that it no longer has to fetch the default values
from the MySQL server when performing inserts, making these
operations faster and more efficient.
Important
Native default value support is enabled for
NDBCLUSTER tables created in
MySQL Cluster NDB 7.1.0 and later. A table created in a
previous version of MySQL Cluster does
not automatically support native default
values after the cluster is upgraded, but instead continues to
use default values supplied by the MySQL server until it is
upgraded using an offline ALTER
TABLE statement.
As part of this change, new methods relating to default values
have been added to the Column and
Table classes in the NDB API. For more
information, see Column::getDefaultValue() ,
Column::setDefaultValue() , and
Table::hasDefaultValues() .
Bugs fixed: Important Change:
The --with-ndb-port-base option for
configure has been removed. It is now handled
as an unknown and invalid option if you attempt to use it when
configuring a build of MySQL Cluster.
(Bug#47941) See also Bug#38502.
The value set by the
--ndb-cluster-connection-pool
option was not shown in the output of
SHOW STATUS LIKE
'NDB%' .
(Bug#44118)
When building storage engines on Windows it was not possible to
specify additional libraries within the CMake file required for
the build. An ${engine}_LIBS macro has been
added to the files to support these additional storage-engine
specific libraries.
(Bug#47797)
When building a pluggable storage engine on Windows, the engine
name could be based on the directory name where the engine was
located, rather than the configured storage engine name.
(Bug#47795)
Installation of MySQL on Windows would fail to set the correct
location for the character set files, which could lead to
mysqld and mysql failing
to initialize properly.
(Bug#17270)
17.7.2. Changes in MySQL Cluster NDB 7.0
This section contains change history information for MySQL Cluster
releases based on version 7.0 of the
NDBCLUSTER storage engine.
Important
Previously, version 7.0 of
NDBCLUSTER was known as version
6.4, and early development versions of MySQL Cluster NDB 7.0
were known as “MySQL Cluster 6.4”. The first four
releases in this series were identified using NDB 6.4.x version
numbers (NDB 6.4.0 through NDB 6.4.3).
MySQL Cluster NDB 7.0.4 is the fifth release in this series. All
future MySQL Cluster NDB 7.0 releases will use NDB 7.0.x version
numbers.
Users running MySQL Cluster NDB 6.4.3 should upgrade to MySQL
Cluster NDB 7.0.4 (or a later MySQL Cluster NDB 7.0 release).
For an overview of new features added in MySQL Cluster NDB 7.0,
see Section 17.1.4.5, “MySQL Cluster Development in MySQL Cluster NDB 7.0”.
When upgrading to a MySQL Cluster NDB 7.0 release from earlier
MySQL Cluster releases, you should be aware of the following
issues:
To perform an online upgrade to MySQL Cluster NDB 7.0 from a
MySQL Cluster NDB 6.3 release, you must upgrade to MySQL
Cluster NDB 7.0.6 or later. See
Section 17.2.6.2, “MySQL Cluster 5.1 and MySQL Cluster NDB 6.x/7.x Upgrade and Downgrade
Compatibility”,
for more information.
You cannot use a mix of MySQL Cluster NDB 7.0 and earlier
binaries, except as part of an online upgrade (where this is
supported). You should replace all existing MySQL Cluster
executables with the MySQL Cluster NDB 7.0 versions.
NDB API applications built against previous MySQL Cluster
versions (NDB 6.3 and earlier) must be recompiled against
against the MySQL Cluster NDB 7.0 sources.
17.7.2.1. Changes in MySQL Cluster NDB 7.0.13 (5.1.41-ndb-7.0.13) (Not yet released)
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.12.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.41 (see
Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
When using NoOfReplicas equal to 1 or 2, if
data nodes from one node group were restarted 256 times and
applications were running traffic such that it would encounter
NDB error 1204
(Temporary failure, distribution
changed), the live node in the node group would
crash, causing the cluster to crash as well. The crash occurred
only when the error was encountered on the 256th restart; having
the error on any previous or subsequent restart did not cause
any problems.
(Bug#50930)
17.7.2.2. Changes in MySQL Cluster NDB 7.0.12 (5.1.41-ndb-7.0.12) (03 February 2010)
This is a testing release only, for the purpose of testing the
fix for Bug#49190. This release is otherwise identical to MySQL
Cluster NDB 7.0.11; users of MySQL Cluster 7.0.11 who are not
interested in testing this change should wait to upgrade to
MySQL Cluster NDB 7.0.13 when that version of MySQL Cluster
becomes available.
Obtaining MySQL Cluster NDB 7.0.12.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.41 (see
Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
ndbmtd started on a single-core machine could
sometimes fail with a Job Buffer Full
error when MaxNoOfExecutionThreads was set
greater than LockExecuteThreadToCPU . Now a
warning is logged when this occurs.
(Bug#50582)
When a primary key lookup on an NDB
table containing one or more BLOB
columns was executed in a transaction, a shared lock on any blob
tables used by the NDB table was
held for the duration of the transaction. (This did not occur
for indexed or non-indexed WHERE conditions.)
Now in such cases, the lock is released after all
BLOB data has been read.
(Bug#49190)
17.7.2.3. Changes in MySQL Cluster NDB 7.0.11 (5.1.41-ndb-7.0.11) (02 February 2010)
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.10.
Obtaining MySQL Cluster NDB 7.0.11.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.39 (see
Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Change:
The maximum allowed value of the
ndb_autoincrement_prefetch_sz
system variable has been increased from 256 to 65536.
(Bug#50621)
Added multi-threaded ordered index building capability during
system restarts or node restarts, controlled by the
BuildIndexThreads data node configuration
parameter (also introduced in this release).
Cluster Replication:
Due to the fact that no timestamp is available for delete
operations, a delete using NDB$MAX() is
actually processed as NDB$OLD . However,
because this is not optimal for some use cases,
NDB$MAX_DELETE_WIN() is added as a conflict
resolution function; if the “timestamp” column
value for a given row adding or updating an existing row coming
from the master is higher than that on the slave, it is applied
(as with NDB$MAX() ); however, delete
operations are treated as always having the higher value.
See
NDB$MAX_DELETE_WIN(column_name ),
for more information.
(Bug#50650)
Bugs fixed:
Initial start of partitioned nodes did not work correctly. This
issue was observed in MySQL Cluster NDB 7.0 only.
(Bug#50661)
The CREATE NODEGROUP client command in
ndb_mgm could sometimes cause the forced
shutdown of a data node.
(Bug#50594)
Local query handler information was not reported or written to
the cluster log correctly. This issue is thought to have been
introduced in MySQL Cluster NDB 7.0.10.
(Bug#50467)
Dropping unique indexes in parallel while they were in use could
cause node and cluster failures.
(Bug#50118)
When attempting to join a running cluster whose management
server had been started with the
--nowait-nodes option and
having SQL nodes with dynamically allocated node IDs, a second
management server failed with spurious INTERNAL
ERROR: Found dynamic ports with value in config...
error messages.
(Bug#49807)
When setting the LockPagesInMainMemory
configuration parameter failed, only the error Failed
to memlock pages... was returned. Now in such cases
the operating system's error code is also returned.
(Bug#49724)
If a query on an NDB table compared
a constant string value to a column, and the length of the
string was greater than that of the column, condition pushdown
did not work correctly. (The string was truncated to fit the
column length before being pushed down.) Now in such cases, the
condition is no longer pushed down.
(Bug#49459)
ndbmtd was not built on Windows
(CMake did not provide a build target for
it).
(Bug#49325)
Performing intensive inserts and deletes in parallel with a high
scan load could a data node crashes due to a failure in the
DBACC kernel block. This was because checking
for when to perform bucket splits or merges considered the first
4 scans only.
(Bug#48700)
During Start Phases 1 and 2, the STATUS
command sometimes (falsely) returned Not
Connected for data nodes running
ndbmtd.
(Bug#47818)
When performing a DELETE that
included a left join from an NDB
table, only the first matching row was deleted.
(Bug#47054)
Under some circumstances, the DBTC kernel
block could send an excessive number of commit and completion
messages which could lead to a the job buffer filling up and
node failure. This was especially likely to occur when using
ndbmtd with a single data node.
(Bug#45989)
When setting LockPagesInMainMemory , the
stated memory was not allocated when the node was started, but
rather only when the memory was used by the data node process
for other reasons.
(Bug#37430)
Trying to insert more rows than would fit into an
NDB table caused data nodes to crash. Now in
such situations, the insert fails gracefully with error 633
Table fragment hash index has reached maximum
possible size.
(Bug#34348)
On Mac OS X or Windows, sending a SIGHUP
signal to the server or an asynchronous flush (triggered by
flush_time ) caused the server
to crash.
(Bug#47525)
The ARCHIVE storage engine lost
records during a bulk insert.
(Bug#46961)
When using the ARCHIVE storage
engine, SHOW TABLE STATUS displayed incorrect
information for Max_data_length ,
Data_length and
Avg_row_length .
(Bug#29203)
17.7.2.4. Changes in MySQL Cluster NDB 7.0.10 (5.1.39-ndb-7.0.10) (15 December 2009)
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.9a.
Obtaining MySQL Cluster NDB 7.0.10.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.39 (see
Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Added the ndb_mgmd
--nowait-nodes option, which
allows a cluster that is configured to use multiple management
servers to be started using fewer than the number configured.
This is most likely to be useful when a cluster is configured
with two management servers and you wish to start the cluster
using only one of them.
See Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”, for more
information.
(Bug#48669)
This enhanced functionality is supported for upgrades from MySQL
Cluster NDB 6.3 when the NDB engine
version is 6.3.29 or later.
(Bug#48528, Bug#49163)
The output from ndb_config
--configinfo
--xml now indicates, for each
configuration parameter, the following restart type information:
(Bug#47366)
Bugs fixed:
Node takeover during a system restart occurs when the REDO log
for one or more data nodes is out of date, so that a node
restart is invoked for that node or those nodes. If this happens
while a mysqld process is attached to the
cluster as an SQL node, the mysqld takes a
global schema lock (a row lock), while trying to set up
cluster-internal replication.
However, this setup process could fail, causing the global
schema lock to be held for an excessive length of time, which
made the node restart hang as well. As a result, the mysqld
failed to set up cluster-internal replication, which led to
tables being read-only, and caused one node to hang during the
restart.
Note
This issue could actually occur in MySQL Cluster NDB 7.0 only,
but the fix was also applied MySQL Cluster NDB 6.3, in order
to keep the two codebases in alignment.
(Bug#49560)
Sending SIGHUP to a mysqld
running with the --ndbcluster and
--log-bin options caused the
process to crash instead of refreshing its log files.
(Bug#49515)
If the master data node receiving a request from a newly-started
API or data node for a node ID died before the request has been
handled, the management server waited (and kept a mutex) until
all handling of this node failure was complete before responding
to any other connections, instead of responding to other
connections as soon as it was informed of the node failure (that
is, it waited until it had received a NF_COMPLETEREP signal
rather than a NODE_FAILREP signal). On visible effect of this
misbehavior was that it caused management client commands such
as SHOW and ALL STATUS to respond with unnecessary slowness in
such circumstances.
(Bug#49207)
Attempting to create more than 11435 tables failed with Error
306 (Out of fragment records in DIH).
(Bug#49156)
When evaluating the options
--include-databases ,
--include-tables ,
--exclude-databases , and
--exclude-tables , the
ndb_restore program overwrote the result of
the database-level options with the result of the table-level
options rather than merging these results together, sometimes
leading to unexpected and unpredictable results.
As part of the fix for this problem, the semantics of these
options have been clarified; because of this, the rules
governing their evaluation have changed slightly. These changes
be summed up as follows:
All --include-* and
--exclude-* options are now evaluated from
right to left in the order in which they are passed to
ndb_restore.
All --include-* and
--exclude-* options are now cumulative.
In the event of a conflict, the first (rightmost) option
takes precedence.
For more detailed information and examples, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#48907)
When performing tasks that generated large amounts of I/O (such
as when using ndb_restore), an internal
memory buffer could overflow, causing data nodes to fail with
signal 6.
Subsequent analysis showed that this buffer was not actually
required, so this fix removes it.
(Bug#48861)
Exhaustion of send buffer memory or long signal memory caused
data nodes to crash. Now an appropriate error message is
provided instead when this situation occurs.
(Bug#48852)
In some situations, when it was not possible for an SQL node to
start a schema transaction (necessary, for instance, as part of
an online ALTER TABLE ),
NDBCLUSTER did not correctly
indicate the error to the MySQL server, which led
mysqld to crash.
(Bug#48841)
Under certain conditions, accounting of the number of free scan
records in the local query handler could be incorrect, so that
during node recovery or a local checkpoint operations, the LQH
could find itself lacking a scan record that is expected to
find, causing the node to crash.
(Bug#48697) See also Bug#48564.
The creation of an ordered index on a table undergoing DDL
operations could cause a data node crash under certain
timing-dependent conditions.
(Bug#48604)
During an LCP master takeover, when the newly elected master did
not receive a COPY_GCI LCP protocol message
but other nodes participating in the local checkpoint had
received one, the new master could use an uninitialized
variable, which caused it to crash.
(Bug#48584)
When running many parallel scans, a local checkpoint (which
performs a scan internally) could find itself not getting a scan
record, which led to a data node crash. Now an extra scan record
is reserved for this purpose, and a problem with obtaining the
scan record returns an appropriate error (error code 489,
Too many active scans).
(Bug#48564)
During a node restart, logging was enabled on a per-fragment
basis as the copying of each fragment was completed but local
checkpoints were not enabled until all fragments were copied,
making it possible to run out of redo log file space
(NDB error code 410) before the
restart was complete. Now logging is enabled only after all
fragments has been copied, just prior to enabling local
checkpoints.
(Bug#48474)
When using very large transactions containing many inserts,
ndbmtd could fail with Signal
11 without an easily detectable reason, due to an
internal variable being unitialized in the event that the
LongMessageBuffer was overloaded. Now, the
variable is initialized in such cases, avoiding the crash, and
an appropriate error message is generated.
(Bug#48441) See also Bug#46914.
A data node crashing while restarting, followed by a system
restart could lead to incorrect handling of redo log metadata,
causing the system restart to fail with Error while
reading REDO log.
(Bug#48436)
Starting a mysqld process with
--ndb-nodeid (either as
a command-line option or by assigning it a value in
my.cnf ) caused the
mysqld to get only the corresponding
connection from the [mysqld] section in the
config.ini file having the matching ID,
even when connection pooling was enabled (that is, when the
mysqld process was started with
--ndb-cluster-connection-pool set
greater than 1).
(Bug#48405) See also Bug#27644, Bug#38590, Bug#41592.
The configuration check that each management server runs to
verify that all connected ndb_mgmd processes
have the same configuration could fail when a configuration
change took place while this check was in progress. Now in such
cases, the configuration check is rescheduled for a later time,
after the change is complete.
(Bug#48143)
When employing NDB native backup to
back up and restore an empty NDB
table that used a non-sequential
AUTO_INCREMENT value, the
AUTO_INCREMENT value was not restored
correctly.
(Bug#48005)
ndb_config --xml
--configinfo now indicates that
parameters belonging in the [SCI] ,
[SCI DEFAULT] , [SHM] , and
[SHM DEFAULT] sections of the
config.ini file are deprecated or
experimental, as appropriate.
(Bug#47365)
NDB stores blob column data in a
separate, hidden table that is not accessible from MySQL. If
this table was missing for some reason (such as accidental
deletion of the file corresponding to the hidden table) when
making a MySQL Cluster native backup, ndb_restore crashed when
attempting to restore the backup. Now in such cases, ndb_restore
fails with the error message Table
table_name has blob column
(column_name ) with missing parts
table in backup instead.
(Bug#47289)
In MySQL Cluster NDB 7.0, ndb_config and
ndb_error_reporter were printing warnings
about management and data nodes running on the same host to
stdout instead of stderr ,
as was the case in earlier MySQL Cluster release series.
(Bug#44689, Bug#49160) See also Bug#25941.
DROP DATABASE failed when there
were stale temporary NDB tables in
the database. This situation could occur if
mysqld crashed during execution of a
DROP TABLE statement after the
table definition had been removed from
NDBCLUSTER but before the
corresponding .ndb file had been removed
from the crashed SQL node's data directory. Now, when
mysqld executes DROP
DATABASE , it checks for these files and removes them
if there are no corresponding table definitions for them found
in NDBCLUSTER .
(Bug#44529)
Creating an NDB table with an
excessive number of large BIT
columns caused the cluster to fail. Now, an attempt to create
such a table is rejected with error 791 (Too many
total bits in bitfields).
(Bug#42046) See also Bug#42047.
When a long-running transaction lasting long enough to cause
Error 410 (REDO log files overloaded) was
later committed or rolled back, it could happen that
NDBCLUSTER was not able to release
the space used for the REDO log, so that the error condition
persisted indefinitely.
The most likely cause of such transactions is a bug in the
application using MySQL Cluster. This fix should handle most
cases where this might occur.
(Bug#36500)
Deprecation and usage information obtained from
ndb_config --configinfo
regarding the PortNumber and
ServerPort configuration parameters was
improved.
(Bug#24584) Disk Data:
When running a write-intensive workload with a very large disk
page buffer cache, CPU usage approached 100% during a local
checkpoint of a cluster containing Disk Data tables.
(Bug#49532) Disk Data:
NDBCLUSTER failed to provide a
valid error message it failed to commit schema transactions
during an initial start if the cluster was configured using the
InitialLogFileGroup parameter.
(Bug#48517) Disk Data:
In certain limited cases, it was possible when the cluster
contained Disk Data tables for ndbmtd to
crash during a system restart.
(Bug#48498) See also Bug#47832. Disk Data:
Repeatedly creating and then dropping Disk Data tables could
eventually lead to data node failures.
(Bug#45794, Bug#48910) Disk Data:
When a crash occurs due to a problem in Disk Data code, the
currently active page list is printed to
stdout (that is, in one or more
ndb_nodeid _out.log
files). One of these lists could contain an endless loop; this
caused a printout that was effectively never-ending. Now in such
cases, a maximum of 512 entries is printed from each list.
(Bug#42431) Disk Data:
When the FileSystemPathUndoFiles
configuration parameter was set to an non-existent path, the
data nodes shut down with the generic error code 2341
(Internal program error). Now in such
cases, the error reported is error 2815 (File not
found).
Cluster Replication:
When expire_logs_days was set,
the thread performing the purge of the log files could deadlock,
causing all binary log operations to stop.
(Bug#49536) Cluster API:
When a DML operation failed due to a uniqueness violation on an
NDB table having more than one
unique index, it was difficult to determine which constraint
caused the failure; it was necessary to obtain an
NdbError object, then decode its
details property, which in could lead to
memory management issues in application code.
To help solve this problem, a new API method
Ndb::getNdbErrorDetail() is added, providing
a well-formatted string containing more precise information
about the index that caused the unque constraint violation. The
following additional changes are also made in the NDB API:
For more information, see
Ndb::getNdbErrorDetail() ,
The NdbError Structure, and
Dictionary::listObjects() .
(Bug#48851) Cluster API:
When using blobs, calling getBlobHandle()
requires the full key to have been set using
equal() , because
getBlobHandle() must access the key for
adding blob table operations. However, if
getBlobHandle() was called without first
setting all parts of the primary key, the application using it
crashed. Now, an appropriate error code is returned instead.
(Bug#28116, Bug#48973)
The mysql_real_connect() C API
function only attempted to connect to the first IP address
returned for a hostname. This could be a problem if a hostname
mapped to multiple IP address and the server was not bound to
the first one returned. Now
mysql_real_connect() attempts to
connect to all IPv4 or IPv6 addresses that a domain name maps
to.
(Bug#45017) See also Bug#47757.
17.7.2.5. Changes in MySQL Cluster NDB 7.0.9b (5.1.39-ndb-7.0.9b) (10 November 2009)
This release includes a fix for Bug#48651, which was discovered
in MySQL Cluster NDB 7.0.9a shortly after release. The MySQL
Cluster NDB 7.0.9b release is identical in all other respects to
MySQL Cluster NDB 7.0.9a. Users who have already installed MySQL
Cluster NDB 7.0.9a should upgrade to MySQL Cluster NDB 7.0.9b as
soon as possible; users seeking to upgrade from any other
previous MySQL Cluster 7.0 release should upgrade to MySQL
Cluster NDB 7.0.9b instead.
Obtaining MySQL Cluster NDB 7.0.9b.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.39 (see
Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.2.6. Changes in MySQL Cluster NDB 7.0.9a (5.1.39-ndb-7.0.9a) (05 November 2009)Important
MySQL Cluster NDB 7.0.9a was pulled shortly after release due
to Bug#48651. Users seeking to upgrade from a previous MySQL
Cluster NDB 7.0 release should instead use MySQL Cluster NDB
7.0.9b, which contains a fix for this bug, in addition to all
bugfixes and improvements made in MySQL Cluster NDB 7.0.9a.
This release includes a fix for Bug#48531, which was discovered
in MySQL Cluster NDB 7.0.9 shortly after release. The MySQL
Cluster NDB 7.0.9a release is identical in all other respects to
MySQL Cluster NDB 7.0.9. Users who have already installed MySQL
Cluster NDB 7.0.9 should upgrade to MySQL Cluster NDB 7.0.9a as
soon as possible; users seeking to upgrade from any other
previous MySQL Cluster 7.0 release should upgrade to MySQL
Cluster NDB 7.0.9a instead.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.39 (see
Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.2.7. Changes in MySQL Cluster NDB 7.0.9 (5.1.39-ndb-7.0.9) (31 October 2009)Important
MySQL Cluster NDB 7.0.9 and 7.0.9a were pulled shortly after
being released due to Bug#48531 and Bug#48651. Users seeking
to upgrade from a previous MySQL Cluster NDB 7.0 release
should instead use MySQL Cluster NDB 7.0.9b, which contains
fixes for these critical bugs, in addition to all bugfixes and
improvements made in MySQL Cluster NDB 7.0.9.
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.8a.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.39 (see
Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Performance:
Significant improvements in redo log handling and other
filesystem operations can yield a considerable reduction in the
time required for restarts. While actual restart times observed
in a production setting will naturally vary according to
database size, hardware, and other conditions, our own
preliminary testing shows that these optimizations can yield
startup times that are faster than those typical of previous
MySQL Cluster releases by a factor of 50 or more.
Bugs fixed: Important Change:
The --with-ndb-port-base option for
configure did not function correctly, and has
been deprecated. Attempting to use this option produces the
warning Ignoring deprecated option
--with-ndb-port-base.
Beginning with MySQL Cluster NDB 7.1.0, the deprecation warning
itself is removed, and the --with-ndb-port-base
option is simply handled as an unknown and invalid option if you
try to use it.
(Bug#47941) See also Bug#38502.
After upgrading a MySQL Cluster containing tables having unique
indexes from an NDB 6.3 release to an NDB 7.0 release, attempts
to create new unique indexes failed with inconsistent
trigger errors (error code
293).
For more information (including a workaround for previous MySQL
Cluster NDB 7.0 releases), see
Section 17.2.6.2, “MySQL Cluster 5.1 and MySQL Cluster NDB 6.x/7.x Upgrade and Downgrade
Compatibility”.
(Bug#48416)
When a data node failed to start due to inability to recreate or
drop objects during schema restoration (for example:
insufficient memory was available to the data node process on
account of issues not directly related to MySQL Cluster on the
host machine), the reason for the failure was not provided. Now
is such cases, a more informative error message is logged.
(Bug#48232)
A table that was created following an upgrade from a MySQL
Cluster NDB 6.3 release to MySQL Cluster NDB 7.0 (starting with
version 6.4.0) or later was dropped by a system restart. This
was due to a change in the format of NDB
schema files and the fact that the upgrade of the format of
existing NDB 6.3 schema files to the
NDB 7.0 format failed to change the version
number contained in the file; this meant that a system restart
re-ran the upgrade routine, which interpreted the newly-created
table as an uncommitted table (which by definition ought not to
be saved). Now the version number of upgraded
NDB 6.3 schema files is set correctly during
the upgrade process.
(Bug#48227)
In certain cases, performing very large inserts on
NDB tables when using
ndbmtd caused the memory allocations for
ordered or unique indexes (or both) to be exceeded. This could
cause aborted transactions and possibly lead to data node
failures.
(Bug#48037) See also Bug#48113.
For UPDATE
IGNORE statements, batching of updates is now
disabled. This is because such statements failed when batching
of updates was employed if any updates violated a unique
constraint, to the fact a unique constraint violation could not
be handled without aborting the transaction.
(Bug#48036)
Starting a data node with a very large amount of
DataMemory (approximately 90G or more) could
lead to crash of the node due to job buffer congestion.
(Bug#47984)
In some cases, ndbmtd could allocate more
space for the undo buffer than was actually available, leading
to a failure in the LGMAN kernel block and
subsequent failure of the data node.
(Bug#47966)
When an UPDATE statement was
issued against an NDB table where
an index was used to identify rows but no data was actually
changed, the NDB storage returned
zero found rows.
For example, consider the table created and populated using
these statements:
CREATE TABLE t1
(
c1 INT NOT NULL,
c2 INT NOT NULL,
PRIMARY KEY(c1),
KEY(c2)
)
ENGINE = NDB;
INSERT INTO t1 VALUES(1, 1);
The following UPDATE statements,
even though they did not change any rows, each still matched a
row, but this was reported incorrectly in both cases, as shown
here:
mysql> UPDATE t1 SET c2 = 1 WHERE c1 = 1;
Query OK, 0 rows affected (0.00 sec)
Rows matched: 0 Changed: 0 Warnings: 0
mysql> UPDATE t1 SET c1 = 1 WHERE c2 = 1;
Query OK, 0 rows affected (0.00 sec)
Rows matched: 0 Changed: 0 Warnings: 0
Now in such cases, the number of rows matched is correct. (In
the case of each of the example
UPDATE statements just shown,
this is displayed as Rows matched: 1, as it should be.)
This issue could affect UPDATE
statements involving any indexed columns in
NDB tables, regardless of the type
of index (including KEY , UNIQUE
KEY , and PRIMARY KEY ) or the number
of columns covered by the index.
(Bug#47955)
On Solaris, shutting down a management node failed when issuing
the command to do so from a client connected to a different
management node.
(Bug#47948)
After changing the value of DiskSyncSize to
4294967039 (the maximum) in the config.ini
file and reloading the cluster configuration, the new value was
displayed in the update information written into the cluster log
as a signed number instead of unsigned.
(Bug#47944) See also Bug#47932.
On Solaris 10 for SPARC, ndb_mgmd failed to
parse the config.ini file when the
DiskSyncSize configuration parameter, whose
permitted range of values is 32768 to 4294967039, was set equal
to 4294967040 (which is also the value of the internal constant
MAX_INT_RNIL ), nor could
DiskSyncSize be set successfully any higher
than the minimum value.
(Bug#47932) See also Bug#47944.
Setting FragmentLogFileSize to a value
greater than 256 MB led to errors when trying to read the redo
log file.
(Bug#47908)
SHOW CREATE TABLE did not display
the AUTO_INCREMENT value for
NDB tables having
AUTO_INCREMENT columns.
(Bug#47865)
An optimization in MySQL Cluster NDB 7.0 causes the
DBDICT kernel block to copy several tables at
a time when synchronizing the data dictionary to a newly-started
node; previously, this was done one table at a time. However,
when NDB tables were sufficiently large and
numerous, the internal buffer for storing them could fill up,
causing a data node crash.
In testing, it was found that having 100 NDB
tables with 128 columns each was enough to trigger this issue.
(Bug#47859)
Under some circumstances, when a scan encountered an error early
in processing by the DBTC kernel block (see
The DBTC Block), a node
could crash as a result. Such errors could be caused by
applications sending incorrect data, or, more rarely, by a
DROP TABLE operation executed in
parallel with a scan.
(Bug#47831)
When starting a node and synchronizing tables, memory pages were
allocated even for empty fragments. In certain situations, this
could lead to insufficient memory.
(Bug#47782)
During an upgrade, newer nodes (NDB kernel block
DBTUP ) could in some cases try to use the
long signal format for communication with older nodes
(DBUTIL kernel block) that did not understand
the newer format, causing older data nodes to fail after
restarting.
(Bug#47740)
A very small race-condition between
NODE_FAILREP and
LQH_TRANSREQ signals when handling node
failure could lead to operations (locks) not being taken over
when they should have been, and subsequently becoming stale.
This could lead to node restart failures, and applications
getting into endless lock-conflicts with operations that were
not released until the node was restarted.
(Bug#47715) See also Bug#41297.
In some cases, the MySQL Server tried to use an error status
whose value had never been set. The problem in the code that
caused this, in hostname.cc , manifested
when using debug builds of mysqld in MySQL
Cluster replication.
This fix brings MySQL Cluster's error handling in
hostname.cc in line with what is
implemented in MySQL 5.4.
(Bug#47548)
configure failed to honor the
--with-zlib-dir option when trying to build
MySQL Cluster from source.
(Bug#47223)
Performing a system restart of the cluster after having
performed a table reorganization which added partitions caused
the cluster to become inconsistent, possibly leading to a forced
shutdown, in either of the following cases:
When a local checkpoint was in progress but had not yet
completed, new partitions were not restored; that is, data
that was supposed to be moved could be lost instead, leading
to an inconsistent cluster. This was due to an issue whereby
the DBDIH kernel block did not save the
new table definition and instead used the old one (the
version having fewer partitions).
When the most recent LCP had completed, ordered indexes and
unlogged tables were still not saved (since these did not
participate in the LCP). In this case, the cluster crashed
during a subsequent system restart, due to the inconsistency
between the main table and the ordered index.
Now, DBDIH is forced in such cases to use the
version of the table definition held by the
DBDICT kernel block, which was (already)
correct and up to date.
(Bug#46585)
ndbd was not built correctly when compiled
using gcc 4.4.0. (The ndbd
binary was built, but could not be started.)
(Bug#46113)
ndb_mgmd failed to close client connections
that had timed out. After running for some time, a race
condition could develop in the management server, due to
ndb_mgmd having exhausted all of its file
descriptors in this fashion.
(Bug#45497) See also Bug#47712.
If a node failed while sending a fragmented long signal, the
receiving node did not free long signal assembly resources that
it had allocated for the fragments of the long signal that had
already been received.
(Bug#44607)
Numeric configuration parameters set in
my.cnf were interpreted as signed rather
than unsigned values. The effect of this was that values of 2G
or more were truncated with the warning [MgmSrvr]
Warning: option 'opt_name ': signed
value opt_value adjusted to
2147483647. Now such parameter values are treated as
unsigned, so that this truncation does not take place.
This issue did not effect parameters set in
config.ini .
(Bug#44448)
When starting a cluster with a great many tables, it was
possible for MySQL client connections as well as the slave SQL
thread to issue DML statements against MySQL Cluster tables
before mysqld had finished connecting to the
cluster and making all tables writeable. This resulted in
Table ... is read only errors for clients
and the Slave SQL thread.
This issue is fixed by introducing the
--ndb-wait-setup option for the
MySQL server. This provides a configurable maximum amount of
time that mysqld waits for all
NDB tables to become writeable,
before allowing MySQL clients or the slave SQL thread to
connect.
(Bug#40679) See also Bug#46955.
When building MySQL Cluster, it was possible to configure the
build using --with-ndb-port without supplying a
port number. Now in such cases, configure
fails with an error.
(Bug#38502) See also Bug#47941.
When the MySQL server SQL mode included
STRICT_TRANS_TABLES , storage
engine warnings and error codes specific to
NDB were returned when errors occurred,
instead of the MySQL server errors and error codes expected by
some programming APIs (such as Connector/J) and applications.
(Bug#35990)
When a copying operation exhausted the available space on a data
node while copying large BLOB
columns, this could lead to failure of the data node and a
Table is full error on the SQL node which
was executing the operation. Examples of such operations could
include an ALTER TABLE that
changed an INT column to a
BLOB column, or a bulk insert of
BLOB data that failed due to
running out of space or to a duplicate key error.
(Bug#34583, Bug#48040) See also Bug#41674, Bug#45768. Replication:
When mysqlbinlog
--verbose was used to read a
binary log that had been recorded using the row-based format,
the output for events that updated some but not all columns of
tables was not correct.
(Bug#47323) Disk Data:
A local checkpoint of an empty fragment could cause a crash
during a system restart which was based on that LCP.
(Bug#47832) See also Bug#41915. Disk Data:
Multi-threaded data nodes could in some cases attempt to access
the same memory structure in parallel, in a non-safe manner.
This could result in data node failure when running
ndbmtd while using Disk Data tables.
(Bug#44195) See also Bug#46507. Cluster Replication:
When using multiple active replication channels, it was
sometimes possible that a node group would fail on the slave
cluster, causing the slave cluster to shut down.
(Bug#47935) Cluster Replication:
When recording a binary log using the
--ndb-log-update-as-write and
--ndb-log-updated-only options
(both enabled by default) and later attempting to apply that
binary log with mysqlbinlog, any operations
that were played back from the log but which updated only some
(but not all) columns caused any columns that were not updated
to be reset to their default values.
(Bug#47674) See also Bug#47323, Bug#46662. Cluster Replication:
mysqlbinlog failed to apply correctly a
binary log that had been recorded using
--ndb-log-update-as-write=1 .
(Bug#46662) See also Bug#47323, Bug#47674. Cluster API:
If an NDB API program reads the same column more than once, it
is possible exceed the maximum allowable message size, in which
case the operation should be aborted due to NDB error 880
Tried to read too much - too many getValue
calls, however due to a change introduced in MySQL
Cluster NDB 6.3.18, the check for this was not done correctly,
which instead caused a data node crash.
(Bug#48266) Cluster API:
The NDB API methods Dictionary::listEvents() ,
Dictionary::listIndexes() ,
Dictionary::listObjects() , and
NdbOperation::getErrorLine() formerly had
both const and non-const
variants. The non-const versions of these
methods have been removed. In addition, the
NdbOperation::getBlobHandle() method has been
re-implemented in order to provide consistent internal
semantics.
(Bug#47798) Cluster API:
A duplicate read of a column caused NDB API applications to
crash.
(Bug#45282) Cluster API:
The error handling shown in the example file
ndbapi_scan.cpp included with the MySQL
Cluster distribution was incorrect.
(Bug#39573)
Installation of MySQL on Windows would fail to set the correct
location for the character set files, which could lead to
mysqld and mysql failing
to initialize properly.
(Bug#17270)
17.7.2.8. Changes in MySQL Cluster NDB 7.0.8a (5.1.37-ndb-7.0.8a) (07 October 2009)
This release includes a fix for Bug#47844, which was discovered
in MySQL Cluster NDB 7.0.8 shortly after release. The MySQL
Cluster NDB 7.0.8a release is identical in all other respects to
MySQL Cluster NDB 7.0.8. Users who have already installed MySQL
Cluster NDB 7.0.8 should upgrade to MySQL Cluster NDB 7.0.8a as
soon as possible; users seeking to upgrade from MySQL Cluster
NDB 7.0.7 or another previous MySQL Cluster 7.0 release should
upgrade to MySQL Cluster NDB 7.0.8a instead.
Obtaining MySQL Cluster NDB 7.0.8a.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.37 (see
Section C.1.11, “Changes in MySQL 5.1.37 (13 July 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
The disconnection of an API or SQL node having a node ID greater
than 49 caused a forced shutdown of the cluster.
(Bug#47844)
The error message text for NDB
error code 410 (REDO log files
overloaded...) was truncated.
(Bug#23662)
17.7.2.9. Changes in MySQL Cluster NDB 7.0.8 (5.1.37-ndb-7.0.8) (30 September 2009)Important
MySQL Cluster NDB 7.0.8 was pulled shortly after release due
to Bug#47844. Users seeking to upgrade from a previous MySQL
Cluster NDB 7.0 release should instead use MySQL Cluster NDB
7.0.8a, which contains a fix for this bug, in addition to all
bugfixes and improvements made in MySQL Cluster NDB 7.0.8.
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.7.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.37 (see
Section C.1.11, “Changes in MySQL 5.1.37 (13 July 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: Important Change:
Previously, the MySQL Cluster management node and data node
programs, when run on Windows platforms, required the
--nodaemon option in order to produce console
output. Now, these programs run in the foreground when invoked
from the command line on Windows, which is the same behavior
that mysqld.exe displays on Windows.
(Bug#45588) Cluster Replication: Important Change:
In a MySQL Cluster acting as a replication slave and having
multiple SQL nodes, only the SQL node receiving events directly
from the master recorded DDL statements in its binary logs
unless this SQL node also had binary logging enabled; otherwise,
other SQL nodes in the slave cluster failed to log DDL
statements, regardless of their individual
--log-bin settings.
The fix for this issue aligns binary logging of DDL statements
with that of DML statements. In particular, you should take note
of the following:
DDL and DML statements on the master cluster are logged with
the server ID of the server that actually writes the log.
DDL and DML statements on the master cluster are logged by
any attached mysqld that has binary
logging enabled.
Replicated DDL and DML statements on the slave are logged by
any attached mysqld that has both
--log-bin and
--log-slave-updates enabled.
Replicated DDL and DML statements are logged with the server
ID of the original (master) MySQL server by any attached
mysqld that has both
--log-bin and
--log-slave-updates enabled.
Affect on upgrades.
When upgrading from a previous MySQL CLuster release, you
should perform either one of the following:
Upgrade servers that are performing binary logging
before those that are not; do not perform any DDL on
“old” SQL nodes until all SQL nodes have
been upgraded.
Make sure that
--log-slave-updates is
enabled on all SQL nodes performing binary logging prior
to the upgrade, so that all DDL is captured.
Note
Logging of DML statements was not affected by this issue.
(Bug#45756)
The following issues with error logs generated by
ndbmtd were addressed:
The version string was sometimes truncated, or even not
shown, depending on the number of threads in use (the more
threads, the worse the problem). Now the version string is
shown in full, as well as the filenames for all tracefiles
(where available).
In the event of a crash, the thread number of the thread
that crashed was not printed. Now this information is
supplied, if available.
(Bug#47629)
mysqld allocated an excessively large buffer
for handling BLOB values due to
overestimating their size. (For each row, enough space was
allocated to accommodate every
BLOB or
TEXT column value in the result
set.) This could adversely affect performance when using tables
containing BLOB or
TEXT columns; in a few extreme
cases, this issue could also cause the host system to run out of
memory unexpectedly.
(Bug#47574) See also Bug#47572, Bug#47573.
NDBCLUSTER uses a dynamically-allocated
buffer to store BLOB or
TEXT column data that is read
from rows in MySQL Cluster tables.
When an instance of the NDBCLUSTER table
handler was recycled (this can happen due to table definition
cache pressure or to operations such as
FLUSH TABLES or
ALTER TABLE ), if the last row
read contained blobs of zero length, the buffer was not freed,
even though the reference to it was lost. This resulted in a
memory leak.
For example, consider the table defined and populated as shown
here:
CREATE TABLE t (a INT PRIMARY KEY, b LONGTEXT) ENGINE=NDB;
INSERT INTO t VALUES (1, REPEAT('F', 20000));
INSERT INTO t VALUES (2, '');
Now execute repeatedly a SELECT
on this table, such that the zero-length
LONGTEXT row is
last, followed by a FLUSH
TABLES statement (which forces the handler object to
be re-used), as shown here:
SELECT a, length(b) FROM bl ORDER BY a;
FLUSH TABLES;
Prior to the fix, this resulted in a memory leak proportional to
the size of the stored
LONGTEXT value
each time these two statements were executed.
(Bug#47573) See also Bug#47572, Bug#47574.
Large transactions involving joins between tables containing
BLOB columns used excessive
memory.
(Bug#47572) See also Bug#47573, Bug#47574.
After an NDB table had an
ALTER ONLINE
TABLE operation performed on it in a MySQL Cluster
running a MySQL Cluster NDB 6.3.x release, it could not be
upgraded online to a MySQL Cluster NDB 7.0.x release. This issue
was detected using MySQL Cluster NDB 6.3.20, but is likely to
effect any MySQL Cluster NDB 6.3.x release supporting online DDL
operations.
(Bug#47542)
When using multi-threaded data nodes (ndbmtd)
with NoOfReplicas set to a value greater than
2, attempting to restart any of the data nodes caused a forced
shutdown of the entire cluster.
(Bug#47530)
A variable was left uninitialized while a data node copied data
from its peers as part of its startup routine; if the starting
node died during this phase, this could lead a crash of the
cluster when the node was later restarted.
(Bug#47505)
Handling of LQH_TRANS_REQ signals was done
incorrectly in DBLQH when the transaction
coordinator failed during a LQH_TRANS_REQ
session. This led to incorrect handling of multiple node
failures, particularly when using ndbmtd.
(Bug#47476)
The NDB kernel's parser (in
ndb/src/common/util/Parser.cpp ) did not
interpret the backslash (“\ ”)
character correctly.
(Bug#47426)
During an online alter table operation, the new table definition
was made available for users during the prepare-phase when it
should only be exposed during and after a commit. This issue
could affect NDB API applications, mysqld processes, or data
node processes.
(Bug#47375)
Aborting an online add column operation (for example, due to
resource problems on a single data node, but not others) could
lead to a forced node shutdown.
(Bug#47364)
Clients attempting to connect to the cluster during shutdown
could sometimes cause the management server to crash.
(Bug#47325)
The size of the table descriptor pool used in the
DBTUP kernel block was incorrect. This could
lead to a data node crash when an LQH sent a
CREATE_TAB_REF signal.
(Bug#47215) See also Bug#44908.
When a data node restarts, it first runs the redo log until
reaching the latest restorable global checkpoint; after this it
scans the remainder of the redo log file, searching for entries
that should be invalidated so they are not used in any
subsequent restarts. (It is possible, for example, if restoring
GCI number 25, that there might be entries belonging to GCI 26
in the redo log.) However, under certain rare conditions, during
the invalidation process, the redo log files themselves were not
always closed while scanning ahead in the redo log. In rare
cases, this could lead to MaxNoOfOpenFiles
being exceeded, causing a the data node to crash.
(Bug#47171)
For very large values of MaxNoOfTables +
MaxNoOfAttributes , the calculation for
StringMemory could overflow when creating
large numbers of tables, leading to NDB error 773
(Out of string memory, please modify StringMemory
config parameter), even when
StringMemory was set to
100 (100 percent).
(Bug#47170)
The default value for the StringMemory
configuration parameter, unlike other MySQL Cluster
configuration parameters, was not set in
ndb/src/mgmsrv/ConfigInfo.cpp .
(Bug#47166)
Signals from a failed API node could be received after an
API_FAILREQ signal (see
Operations and Signals)
has been received from that node, which could result in invalid
states for processing subsequent signals. Now, all pending
signals from a failing API node are processed before any
API_FAILREQ signal is received.
(Bug#47039) See also Bug#44607.
When reloading the management server configuration, only the
last changed parameter was logged.
(Bug#47036)
When using ndbmtd, a parallel
DROP TABLE operation could cause
data nodes to have different views of which tables should be
included in local checkpoints; this discrepancy could lead to a
node failure during an LCP.
(Bug#46873)
Using triggers on NDB tables caused
ndb_autoincrement_prefetch_sz
to be treated as having the NDB kernel's internal default
value (32) and the value for this variable as set on the
cluster's SQL nodes to be ignored.
(Bug#46712)
Now, when started with
--initial
--reload ,
ndb_mgmd tries to copy the configuration of
an existing ndb_mgmd process with a confirmed
configuration. This works only if the configuration files used
by both management nodes are exactly the same.
(Bug#45495, Bug#46488) See also Bug#42015.
On Windows, ndbd
--initial could hang in
an endless loop while attempting to remove directories.
(Bug#45402)
For multi-threaded data nodes, insufficient fragment records
were allocated in the DBDIH NDB kernel block,
which could lead to error 306 when creating many tables; the
number of fragment records allocated did not take into account
the number of LQH instances.
(Bug#44908)
Running an ALTER TABLE statement
while an NDB backup was in progress caused
mysqld to crash.
(Bug#44695)
When performing auto-discovery of tables on individual SQL
nodes, NDBCLUSTER attempted to overwrite
existing MyISAM .frm
files and corrupted them.
Workaround.
In the mysql client, create a new table
(t2 ) with same definition as the corrupted
table (t1 ). Use your system shell or file
manager to rename the old .MYD file to
the new file name (for example, mv t1.MYD
t2.MYD). In the mysql client,
repair the new table, drop the old one, and rename the new
table using the old file name (for example,
RENAME TABLE t2
TO t1 ).
(Bug#42614)
When started with the --initial
and --reload options, if
ndb_mgmd could not find a configuration file
or connect to another management server, it appeared to hang.
Now, when trying to fetch its configuration from another
management node, ndb_mgmd checks and signals
(Trying to get configuration from other
mgmd(s) ) each 30 seconds that it has not yet done so.
(Bug#42015) See also Bug#45495.
Running ndb_restore with the
--print or --print_log option
could cause it to crash.
(Bug#40428, Bug#33040)
An insert on an NDB table was not
always flushed properly before performing a scan. One way in
which this issue could manifest was that
LAST_INSERT_ID() sometimes failed
to return correct values when using a trigger on an
NDB table.
(Bug#38034)
When a data node received a TAKE_OVERTCCONF
signal from the master before that node had received a
NODE_FAILREP , a race condition could in
theory result.
(Bug#37688) See also Bug#25364, Bug#28717.
Some joins on large NDB tables
having TEXT or
BLOB columns could cause
mysqld processes to leak memory. The joins
did not need to reference the
TEXT or
BLOB columns directly for this
issue to occur.
(Bug#36701)
On Mac OS X 10.5, commands entered in the management client
failed and sometimes caused the client to hang, although
management client commands invoked using the
--execute (or
-e ) option from the system shell worked
normally.
For example, the following command failed with an error and hung
until killed manually, as shown here:
ndb_mgm> SHOW
Warning, event thread startup failed, degraded printouts as result, errno=36
^C
However, the same management client command, invoked from the
system shell as shown here, worked correctly:
shell> ndb_mgm -e "SHOW"
(Bug#35751) See also Bug#34438. Replication:
In some cases, a STOP SLAVE
statement could cause the replication slave to crash. This issue
was specific to MySQL on Windows or Macintosh platforms.
(Bug#45238, Bug#45242, Bug#45243, Bug#46013, Bug#46014, Bug#46030) See also Bug#40796. Disk Data:
Calculation of free space for Disk Data table fragments was
sometimes done incorrectly. This could lead to unnecessary
allocation of new extents even when sufficient space was
available in existing ones for inserted data. In some cases,
this might also lead to crashes when restarting data nodes.
Note
This miscalculation was not reflected in the contents of the
INFORMATION_SCHEMA.FILES table,
as it applied to extents allocated to a fragment, and not to a
file.
(Bug#47072) Cluster API:
In some circumstances, if an API node encountered a data node
failure between the creation of a transaction and the start of a
scan using that transaction, then any subsequent calls to
startTransaction() and
closeTransaction() could cause the same
transaction to be started and closed repeatedly.
(Bug#47329) Cluster API:
Performing multiple operations using the same primary key within
the same
NdbTransaction::execute()
call could lead to a data node crash.
Note
This fix does not make change the fact that performing
multiple operations using the same primary key within the same
execute() is not supported; because there
is no way to determine the order of such operations, the
result of such combined operations remains undefined.
(Bug#44065) See also Bug#44015. API:
The fix for Bug#24507 could lead in some cases to client
application failures due to a race condition. Now the server
waits for the “dummy” thread to return before
exiting, thus making sure that only one thread can initialize
the POSIX threads library.
(Bug#42850)
17.7.2.10. Changes in MySQL Cluster NDB 7.0.7 (5.1.35-ndb-7.0.7) (26 August 2009)
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.6.
Obtaining MySQL Cluster NDB 7.0.7.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.35 (see
Section C.1.13, “Changes in MySQL 5.1.35 (13 May 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Change:
The default value of the DiskIOThreadPool
data node configuration parameter has changed from 8 to 2.
On Solaris platforms, the MySQL Cluster management server and
NDB API applications now use CLOCK_REALTIME
as the default clock.
(Bug#46183)
Formerly, node IDs were represented in the cluster log using a
complex hexadecimal/binary encoding scheme. Now, node IDs are
reported in the cluster log using numbers in standard decimal
notation.
(Bug#44248)
A new option --exclude-missing-columns has been
added for the ndb_restore program. In the
event that any tables in the database or databases being
restored to have fewer columns than the same-named tables in the
backup, the extra columns in the backup's version of the
tables are ignored. For more information, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#43139) Note
This issue, originally resolved in MySQL 5.1.16, re-occurred
due to a later (unrelated) change. The fix has been
re-applied.
(Bug#25984)
Previously, it was possible to disable arbitration only by
setting ArbitrationRank to 0 on all
management and API nodes. A new data node configuration
parameter Arbitration simplifies this task;
to disable arbitration, you can now use Arbitration =
Disabled in the [ndbd default]
section of the config.ini file.
It is now also possible to configure arbitration in such a way
that the cluster waits until the time determined by
ArbitrationTimeout passes for an external
manager to perform arbitration instead of handling it
internally. This can be done by setting Arbitration =
WaitExternal in the [ndbd default]
section of the config.ini file.
The default value for the Arbitration parameter is
Default , which allows arbitration to proceed
normally, as determined by the
ArbitrationRank settings for the management
and API nodes.
For more information, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
Bugs fixed: Packaging:
The pkg installer for MySQL Cluster on
Solaris did not perform a complete installation due to an
invalid directory reference in the post-install script.
(Bug#41998)
The output from ndb_config
--configinfo
--xml
contained quote characters (" ) within quoted
XML attributes, causing it to be not well-formed.
(Bug#46891)
When using multi-threaded data node processes
(ndbmtd), it was possible for LQH threads to
continue running even after all NDB
tables had been dropped. This meant that dropping the last
remaining NDB table during a local
checkpoint could cause multi-threaded data nodes to fail.
(Bug#46890)
During a global checkpoint, LQH threads could run unevenly,
causing a circular buffer oveflow by the Subscription Manager,
which led to data node failure.
(Bug#46782) See also Bug#46123, Bug#46723, Bug#45612.
Restarting the cluster following a local checkpoint and an
online ALTER TABLE on a non-empty
table caused data nodes to crash.
(Bug#46651)
A combination of index creation and drop operations (or creating
and dropping tables having indexes) with node and system
restarts could lead to a crash.
(Bug#46552)
Following an upgrade from MySQL Cluster NDB 6.3.x to MySQL
Cluster NDB 7.0.6, DDL and backup operations failed.
(Bug#46494, Bug#46563)
Full table scans failed to execute when the cluster contained
more than 21 table fragments.
Note
The number of table fragments in the cluster can be calculated
as the number of data nodes, times 8 (that is, times the value
of the internal constant
MAX_FRAG_PER_NODE ), divided by the number
of replicas. Thus, when NoOfReplicas = 1 at
least 3 data nodes were required to trigger this issue, and
when NoOfReplicas = 2 at least 4 data nodes
were required to do so.
(Bug#46490)
Killing MySQL Cluster nodes immediately following a local
checkpoint could lead to a crash of the cluster when later
attempting to perform a system restart.
The exact sequence of events causing this issue was as follows:
Local checkpoint occurs.
Immediately following the LCP, kill the master data node.
Kill the remaining data nodes within a few seconds of
killing the master.
Attempt to restart the cluster.
(Bug#46412)
Creating an index when the cluster had run out of table records
could cause data nodes to crash.
(Bug#46295)
Ending a line in the config.ini file with
an extra semicolon character (; ) caused
reading the file to fail with a parsing error.
(Bug#46242)
When combining an index scan and a delete with a primary key
delete, the index scan and delete failed to initialize a flag
properly. This could in rare circumstances cause a data node to
crash.
(Bug#46069)
OPTIMIZE TABLE on an
NDB table could in some cases cause
SQL and data nodes to crash. This issue was observed with both
ndbd and ndbmtd.
(Bug#45971)
The AutoReconnect configuration parameter for
API nodes (including SQL nodes) has been added. This is intended
to prevent API nodes from re-using allocated node IDs during
cluster restarts. For more information, see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
This fix also introduces two new methods of the
Ndb_cluster_connection class in the NDB API.
For more information, see
Ndb_cluster_connection::set_auto_reconnect() ,
and
Ndb_cluster_connection::get_auto_reconnect() .
(Bug#45921)
DML statements run during an upgrade from MySQL Cluster NDB 6.3
to NDB 7.0 were not handled correctly.
(Bug#45917)
On Windows, the internal
basestring_vsprintf() function did not return
a POSIX-compliant value as expected, causing the management
server to crash when trying to start a MySQL Cluster with more
than 4 data nodes.
(Bug#45733)
The signals used by ndb_restore to send
progress information about backups to the cluster log accessed
the cluster transporter without using any locks. Because of
this, it was theoretically possible that these signals could be
interefered with by heartbeat signals if both were sent at the
same time, causing the ndb_restore messages
to be corrupted.
(Bug#45646)
Due to changes in the way that
NDBCLUSTER handles schema changes
(implementation of schema transactions) in MySQL Cluster NDB
7.0, it was not possible to create MySQL Cluster tables having
more than 16 indexes using a single CREATE
TABLE statement.
This issue occurs only in MySQL Cluster NDB 7.0 releases prior
to 7.0.7 (including releases numbered NDB 6.4.x).
If you are not yet able to upgrade from an earlier MySQL Cluster
NDB 7.0 release, you can work around this problem by creating
the table without any indexes, then adding the indexes using a
separate CREATE INDEX statement
for each index.
(Bug#45525)
storage/ndb/src/common/util/CMakeLists.txt
did not build the BaseString-t test program
for Windows as the equivalent
storage/ndb/src/common/util/Makefile.am
does when building MySQL Cluster on Unix platforms.
(Bug#45099)
Problems could arise when using
VARCHAR columns
whose size was greater than 341 characters and which used the
utf8_unicode_ci collation. In some cases,
this combination of conditions could cause certain queries and
OPTIMIZE TABLE statements to
crash mysqld.
(Bug#45053)
The warning message Possible bug in
Dbdih::execBLOCK_COMMIT_ORD ... could sometimes
appear in the cluster log. This warning is obsolete, and has
been removed.
(Bug#44563)
Debugging code causing ndbd to use file
compression on NTFS filesystems failed with an error. (The code
was removed.) This issue affected debug builds of MySQL Cluster
on Windows platforms only.
(Bug#44418)
ALTER TABLE
REORGANIZE PARTITION could fail with Error 741
(Unsupported alter table) if the
appropriate hash-map was not present. This could occur when
adding nodes online; for example, when going from 2 data nodes
to 3 data nodes with NoOfReplicas=1 , or from
4 data nodes to 6 data nodes with
NoOfReplicas=2 .
(Bug#44301)
Previously, a GCP STOP event was written to
the cluster log as an INFO event. Now it is
logged as a WARNING event instead.
(Bug#43853)
In some cases, OPTIMIZE TABLE on
an NDB table did not free any
DataMemory .
(Bug#43683)
If the cluster crashed during the execution of a
CREATE LOGFILE GROUP statement,
the cluster could not be restarted afterwards.
(Bug#36702) See also Bug#34102. Disk Data: Partitioning:
An NDBCLUSTER table created with a
very large value for the MAX_ROWS option
could — if this table was dropped and a new table with
fewer partitions, but having the same table ID, was created
— cause ndbd to crash when performing a
system restart. This was because the server attempted to examine
each partition whether or not it actually existed.
(Bug#45154) Disk Data:
If the value set in the config.ini file for
FileSystemPathDD ,
FileSystemPathDataFiles , or
FileSystemPathUndoFiles was identical to the
value set for FileSystemPath , that parameter
was ignored when starting the data node with
--initial option. As a result, the Disk Data
files in the corresponding directory were not removed when
performing an initial start of the affected data node or data
nodes.
(Bug#46243)
For an IPv6-enabled MySQL server, privileges specified using
standard IPv4 addresses for hosts were not matched (only
IPv4-mapped addresses were handled correctly).
As part of the fix for this bug, a new build option
--disable-ipv6 has been introduced. Compiling
MySQL with this option causes all IPv6-specific code in the
server to be ignored.
Important
If the server has been compiled using
--disable-ipv6 , it is not able to resolve
hostnames correctly when run in an IPv6 environment.
(Bug#45606) See also Bug#38247, Bug#43006, Bug#45283, Bug#45584.
The hostname cache failed to work correctly.
(Bug#45584) See also Bug#38247, Bug#43006, Bug#45283, Bug#45606.
The number of connection errors from a given host as counted by
the server was periodically reset, with the result that
max_connect_errors was never
reached and invalid hosts were never blocked from trying to
connect.
(Bug#45283) See also Bug#38247, Bug#43006, Bug#45584, Bug#45606.
The IPv6 loopback address ::1 was interpeted
as a hostname rather than a numeric IP address.
In addition, the IPv6-enabled server on Windows interpeted the
hostname localhost as ::1
only, which failed to match the default
root@127.0.0.1 entry in the
mysql.user privilege table.
(Bug#43006) See also Bug#38247, Bug#45283, Bug#45584, Bug#45606.
An IPv6-enabled MySQL server did not resolve the IP addresses of
incoming connections correctly, with the result that a
connection that attempted to match any privilege table entries
using fully-qualified domain names for hostnames or hostnames
using wildcards were dropped.
(Bug#38247) See also Bug#43006, Bug#45283, Bug#45584, Bug#45606.
17.7.2.11. Changes in MySQL Cluster NDB 7.0.6 (5.1.34-ndb-7.0.6) (26 May 2009)
This release incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.5.
Obtaining MySQL Cluster NDB 7.0.5.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 7.0 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.34 (see
Section C.1.15, “Changes in MySQL 5.1.34 (02 April 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
The ndb_config utility program can now
provide an offline dump of all MySQL Cluster configuration
parameters including information such as default and permitted
values, brief description, and applicable section of the
config.ini file. A dump in text format is
produced when running ndb_config with the new
--configinfo option, and in XML format when the
options --configinfo --xml are used together.
For more information and examples, see
Section 17.4.6, “ndb_config — Extract MySQL Cluster Configuration Information”.
Bugs fixed: Important Change: Partitioning:
User-defined partitioning of an
NDBCLUSTER table without any
primary key sometimes failed, and could cause
mysqld to crash.
Now, if you wish to create an
NDBCLUSTER table with user-defined
partitioning, the table must have an explicit primary key, and
all columns listed in the partitioning expression must be part
of the primary key. The hidden primary key used by the
NDBCLUSTER storage engine is not
sufficient for this purpose. However, if the list of columns is
empty (that is, the table is defined using PARTITION BY
[LINEAR] KEY() ), then no explicit primary key is
required.
This change does not effect the partitioning of tables using any
storage engine other than
NDBCLUSTER .
(Bug#40709) Important Change:
Previously, the configuration parameter
NoOfReplicas had no default value. Now the
default for NoOfReplicas is 2, which is the
recommended value in most settings.
(Bug#44746) Important Note:
It was not possible to perform an online upgrade from any MySQL
Cluster NDB 6.x release to MySQL Cluster NDB 7.0.5 or any to
earlier MySQL Cluster NDB 7.0 release.
With this fix, it is possible in MySQL Cluster NDB 7.0.6 and
later to perform online upgrades from MySQL Cluster NDB 6.3.8
and later MySQL Cluster NDB 6.3 releases, or from MySQL Cluster
NDB 7.0.5 or later MySQL Cluster NDB 7.0 releases. Online
upgrades to MySQL Cluster NDB 7.0 releases previous to MySQL
Cluster NDB 7.0.6 from earlier MySQL Cluster releases remain
unsupported; online upgrades from MySQL Cluster NDB 7.0 releases
previous to MySQL Cluster NDB 7.0.5 (including NDB 6.4.x beta
releases) to later MySQL Cluster NDB 7.0 releases also remain
unsupported.
(Bug#44294)
An internal NDB API buffer was not properly initialized.
(Bug#44977)
When a data node had written its GCI marker to the first page of
a megabyte, and that node was later killed during restart after
having processed that page (marker) but before completing a LCP,
the data node could fail with filesystem errors.
(Bug#44952) See also Bug#42564, Bug#44291.
When restarting a data nodes, management and API nodes
reconnecting to it failed to re-use existing ports that had
already been dynamically allocated for communications with that
data node.
(Bug#44866)
When ndb_config could not find the file
referenced by the --config-file option, it
tried to read my.cnf instead, then failed
with a misleading error message.
(Bug#44846)
When a data node was down so long that its most recent local
checkpoint depended on a global checkpoint that was no longer
restorable, it was possible for it to be unable to use optimized
node recovery when being restarted later.
(Bug#44844) See also Bug#26913.
Online upgrades to MySQL Cluster NDB 7.0 from a MySQL Cluster
NDB 6.3 release could fail due to changes in the handling of key
lengths and unique indexes during node recovery.
(Bug#44827)
ndb_config
--xml
did not output any entries for the HostName
parameter. In addition, the default listed for
MaxNoOfFiles was outside the allowed range of
values.
(Bug#44749) See also Bug#44685, Bug#44746.
The output of ndb_config
--xml
did not provide information about all sections of the
configuration file.
(Bug#44685) See also Bug#44746, Bug#44749.
Use of __builtin_expect() had the side effect
that compiler warnings about misuse of =
(assignment) instead of == in comparisons
were lost when building in debug mode. This is no longer
employed when configuring the build with the
--with-debug option.
(Bug#44570) See also Bug#44567.
Inspection of the code revealed that several assignment
operators (= ) were used in place of
comparison operators (== ) in
DbdihMain.cpp .
(Bug#44567) See also Bug#44570.
When using large numbers of configuration parameters, the
management server took an excessive amount of time (several
minutes or more) to load these from the configuration cache when
starting. This problem occurred when there were more than 32
configuration parameters specified, and became progressively
worse with each additional multiple of 32 configuration
parameters.
(Bug#44488)
Building the MySQL Cluster NDB 7.0 tree failed when using the
icc compiler.
(Bug#44310)
SSL connections to SQL nodes failed on big-endian platforms.
(Bug#44295)
Signals providing node state information
(NODE_STATE_REP and
CHANGE_NODE_STATE_REQ ) were not propagated to
all blocks of ndbmtd. This could cause the
following problems:
Inconsistent redo logs when performing a graceful shutdown;
Data nodes crashing when later restarting the cluster, data
nodes needing to perform node recovery during the system
restart, or both.
(Bug#44291) See also Bug#42564.
An NDB internal timing function did not work correctly on
Windows and could cause mysqld to fail on
some AMD processors, or when running inside a virtual machine.
(Bug#44276)
It was possible for NDB API applications to insert corrupt data
into the database, which could subquently lead to data node
crashes. Now, stricter checking is enforced on input data for
inserts and updates.
(Bug#44132)
ndb_restore failed when trying to restore
data on a big-endian machine from a backup file created on a
little-endian machine.
(Bug#44069)
Repeated starting and stopping of data nodes could cause
ndb_mgmd to fail. This issue was observed on Solaris/SPARC.
(Bug#43974)
A number of incorrectly formatted output strings in the source
code caused compiler warnings.
(Bug#43878)
When trying to use a data node with an older version of the
management server, the data node crashed on startup.
(Bug#43699)
In some cases, data node restarts during a system restart could
fail due to insufficient redo log space.
(Bug#43156)
NDBCLUSTER did not build correctly
on Solaris 9 platforms.
(Bug#39080) See also Bug#39036, Bug#39038.
The output of ndbd --help
did not provide clear information about the program's
--initial and --initial-start
options.
(Bug#28905)
It was theoretically possible for the value of a nonexistent
column to be read as NULL , rather than
causing an error.
(Bug#27843) Disk Data:
During a checkpoint, restore points are created for both the
on-disk and in-memory parts of a Disk Data table. Under certain
rare conditions, the in-memory restore point could include or
exclude a row that should have been in the snapshot. This would
later lead to a crash during or following recovery.
This issue was somewhat more likely to be encountered when using
ndbmtd.
(Bug#41915) See also Bug#47832. Disk Data:
This fix supercedes and improves on an earlier fix made for this
bug in MySQL 5.1.18.
(Bug#24521) Cluster Replication:
A failure when setting up replication events could lead to
subsequent data node failures.
(Bug#44915)
17.7.2.12. Changes in MySQL Cluster NDB 7.0.5 (5.1.32-ndb-7.0.5) (20 April 2009 General Availability)
This General Availability (GA) release
incorporates new features in the
NDBCLUSTER storage engine and fixes
recently discovered bugs in MySQL Cluster NDB 7.0.4.
Obtaining MySQL Cluster NDB 7.0.5.
The latest MySQL Cluster NDB 7.0 binaries for supported
platforms can be obtained from
http://dev.mysql.com/downloads/select.php?id=14. Source code for
the latest MySQL Cluster NDB 7.0 release can be obtained from
the same location. You can also access the MySQL Cluster NDB
7.0 development source tree at
https://code.launchpad.net/~mysql/mysql-server/mysql-cluster-7.0.
This release also incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 6.4 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.32 (see
Section C.1.17, “Changes in MySQL 5.1.32 (14 February 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Two new server status variables
Ndb_scan_count and
Ndb_pruned_scan_count have
been introduced.
Ndb_scan_count gives the
number of scans executed since the cluster was last started.
Ndb_pruned_scan_count gives
the number of scans for which
NDBCLUSTER was able to use
partition pruning. Together, these variables can be used to help
determine in the MySQL server whether table scans are pruned by
NDBCLUSTER .
(Bug#44153)
Bugs fixed: Important Note:
Due to problem discovered after the code freeze for this
release, it is not possible to perform an online upgrade from
any MySQL Cluster NDB 6.x release to MySQL Cluster NDB 7.0.5 or
any earlier MySQL Cluster NDB 7.0 release.
This issue is fixed in MySQL Cluster NDB 7.0.6 and later for
upgrades from MySQL Cluster NDB 6.3.8 and later MySQL Cluster
NDB 6.3 releases, or from MySQL Cluster NDB 7.0.5.
(Bug#44294) Cluster Replication:
If data node failed during an event creation operation, there
was a slight risk that a surviving data node could send an
invalid table reference back to NDB, causing the operation to
fail with a false Error 723 (No such
table). This could take place when a data node
failed as a mysqld process was setting up
MySQL Cluster Replication.
(Bug#43754) Cluster API:
The following issues occurred when performing an online
(rolling) upgrade of a cluster to a version of MySQL Cluster
that supports configuration caching from a version that does
not:
When using multiple management servers, after upgrading and
restarting one ndb_mgmd, any remaining
management servers using the previous version of
ndb_mgmd could not synchronize their
configuration data.
The MGM API ndb_mgm_get_configuration()
function failed to obtain configuration data.
(Bug#43641)
If the number of fragments per table rises above a certain
threshold, the DBDIH kernel block's
on-disk table-definition grows large enough to occupy 2 pages.
However, in MySQL Cluster NDB 7.0 (including MySQL Cluster NDB
6.4 releases), only 1 page was actually written, causing table
definitions stored on disk to be incomplete.
This issue was not observed in MySQL Cluster release series
prior to MySQL Cluster NDB 7.0.
(Bug#44135)
TransactionDeadlockDetectionTimeout values
less than 100 were treated as 100. This could cause scans to
time out unexpectedly.
(Bug#44099)
The file ndberror.c contained a C++-style
comment, which caused builds to fail with some C compilers.
(Bug#44036)
A race condition could occur when a data node failed to restart
just before being included in the next global checkpoint. This
could cause other data nodes to fail.
(Bug#43888)
The setting for
ndb_use_transactions was
ignored. This issue was only known to occur in MySQL Cluster NDB
6.4.3 and MySQL Cluster NDB 7.0.4.
(Bug#43236)
When a data node process had been killed after allocating a node
ID, but before making contact with any other data node
processes, it was not possible to restart it due to a node ID
allocation failure.
This issue could effect either ndbd or
ndbmtd processes.
(Bug#43224) This regression was introduced by Bug#42973.
ndb_restore crashed when trying to restore a
backup made to a MySQL Cluster running on a platform having
different endianness from that on which the original backup was
taken.
(Bug#39540)
PID files for the data and management node daemons were not
removed following a normal shutdown.
(Bug#37225)
ndb_restore --print_data did
not handle DECIMAL columns
correctly.
(Bug#37171)
Invoking the management client START BACKUP
command from the system shell (for example, as ndb_mgm
-e "START BACKUP") did not work correctly, unless the
backup ID was included when the command was invoked.
Now, the backup ID is no longer required in such cases, and the
backup ID that is automatically generated is printed to stdout,
similar to how this is done when invoking START
BACKUP within the management client.
(Bug#31754)
When aborting an operation involving both an insert and a
delete, the insert and delete were aborted separately. This was
because the transaction coordinator did not know that the
operations affected on same row, and, in the case of a
committed-read (tuple or index) scan, the abort of the insert
was performed first, then the row was examined after the insert
was aborted but before the delete was aborted. In some cases,
this would leave the row in a inconsistent state. This could
occur when a local checkpoint was performed during a backup.
This issue did not affect primary ley operations or scans that
used locks (these are serialized).
After this fix, for ordered indexes, all operations that follow
the operation to be aborted are now also aborted.
Disk Data:
When using multi-threaded data nodes, DROP
TABLE statements on Disk Data tables could hang.
(Bug#43825) Disk Data:
This fix completes one that was made for this issue in MySQL
Cluster NDB-7.0.4, which did not rectify the problem in all
cases.
(Bug#43632) Cluster Replication:
When creating or altering a table an
NdbEventOperation is created by the
mysqld process to monitor the table for
subsequent logging in the binlog. If this happened during a node
restart there was a chance that the reference count on this
event operation object could be incorrect, which could lead to
an assert in debug MySQL Cluster builds.
(Bug#43752) Cluster API:
If the largest offset of a
RecordSpecification
used for an
NdbRecord
object was for the NULL bits (and thus not a
column), this offset was not taken into account when calculating
the size used for the RecordSpecification .
This meant that the space for the NULL bits
could be overwritten by key or other information.
(Bug#43891) Cluster API:
BIT columns created using the
native NDB API format that were not created as nullable could
still sometimes be overwritten, or cause other columns to be
overwritten.
This issue did not effect tables having
BIT columns created using the
mysqld format (always used by MySQL Cluster SQL nodes).
(Bug#43802)
17.7.2.13. Changes in MySQL Cluster NDB 7.0.4 (5.1.32-ndb-7.0.4) (18 March 2008)
This is a new Beta development release, incorporating new
features in the NDBCLUSTER storage
engine and fixing recently discovered bugs in MySQL Cluster NDB
6.4.3. All feature additions and bugfixes that were made in
MySQL Cluster releases having NDB 6.4.x release numbers are
included in MySQL Cluster 7.0.4.
Important
MySQL Cluster NDB 7.0.4 is the successor to MySQL Cluster NDB
6.4.3. Users running MySQL Cluster NDB 6.4.3 should upgrade to
MySQL Cluster NDB 7.0.4 or a later 7.0.x release.
Obtaining MySQL Cluster NDB 7.0.4.
MySQL Cluster NDB 7.0.4 is a source-only release. You can
obtain the source code from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.32-ndb-7.0.4/.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 6.4 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.32 (see
Section C.1.17, “Changes in MySQL 5.1.32 (14 February 2009)”).
Note
This Beta release, as any other pre-production release, should
not be installed on production level
systems or systems with critical data. Please refer to our bug
database at http://bugs.mysql.com/ for more
details about the individual bugs fixed in this version.
Functionality added or changed: Important Change:
The default values for a number of MySQL Cluster configuration
parameters relating to memory usage and buffering have changed.
These parameters include RedoBuffer ,
LongMessageBuffer ,
BackupMemory ,
BackupDataBufferSize ,
BackupLogBufferSize ,
BackupWriteSize ,
BackupMaxWriteSize ,
SendBufferMemory (when applied to TCP
transporters), and ReceiveBufferMemory .
For more information, see
Section 17.3, “MySQL Cluster Configuration”.
When restoring from backup, ndb_restore now
reports the last global checkpoint reached when the backup was
taken.
(Bug#37384)
Bugs fixed: Cluster API:
Partition pruning did not work correctly for queries involving
multiple range scans.
As part of the fix for this issue, several improvements have
been made in the NDB API, including the addition of a new
NdbScanOperation::getPruned() method, a new
variant of NdbIndexScanOperation::setBound() ,
and a new Ndb::PartitionSpec data structure.
For more information about these changes, see
NdbScanOperation::getPruned() ,
NdbIndexScanOperation::setBound , and
The PartitionSpec Structure.
(Bug#37934)
TimeBetweenLocalCheckpoints was measured from
the end of one local checkpoint to the beginning of the next,
rather than from the beginning of one LCP to the beginning of
the next. This meant that the time spent performing the LCP was
not taken into account when determining the
TimeBetweenLocalCheckpoints interval, so that
LCPs were not started often enough, possibly causing data nodes
to run out of redo log space prematurely.
(Bug#43567)
The management server failed to start correctly in daemon mode.
(Bug#43559)
Following a DROP NODEGROUP command, the
output of SHOW in the
ndb_mgm cliently was not updated to reflect
the fact that the data nodes affected by this command were no
longer part of a node group.
(Bug#43413)
Using indexes containing variable-sized columns could lead to
internal errors when the indexes were being built.
(Bug#43226)
When using ndbmtd, multiple data node
failures caused the remaining data nodes to fail as well.
(Bug#43109)
It was not possible to add new data nodes to the cluster online
using multi-threaded data node processes
(ndbmtd).
(Bug#43108)
Some queries using combinations of logical and comparison
operators on an indexed column in the WHERE
clause could fail with the error Got error 4541
'IndexBound has no bound information' from
NDBCLUSTER.
(Bug#42857) Disk Data:
When using multi-threaded data nodes, dropping a Disk Data table
followed by a data node restart led to a crash.
(Bug#43632) Disk Data:
When using ndbmtd, repeated high-volume
inserts (on the order of 10000 rows inserted at a time) on a
Disk Data table would eventually lead to a data node crash.
(Bug#41398) Disk Data:
When a log file group had an undo log file whose size was too
small, restarting data nodes failed with Read
underflow errors.
As a result of this fix, the minimum allowed
INTIAL_SIZE for an undo log file is now
1M (1 megabyte).
(Bug#29574) Cluster API:
The default NdbRecord structures created by
NdbDictionary could have overlapping null
bits and data fields.
(Bug#43590) Cluster API:
When performing insert or write operations,
NdbRecord allows key columns to be specified
in both the key record and in the attribute record. Only one key
column value for each key column should be sent to the NDB
kernel, but this was not guaranteed. This is now ensured as
follows: For insert and write operations, key column values are
taken from the key record; for scan takeover update operations,
key column values are taken from the attribute record.
(Bug#42238) Cluster API:
Ordered index scans using NdbRecord formerly
expressed a BoundEQ range as separate lower
and upper bounds, resulting in 2 copies of the column values
being sent to the NDB kernel.
Now, when a range is specified by
NdbScanOperation::setBound() , the passed
pointers, key lengths, and inclusive bits are compared, and only
one copy of the equal key columns is sent to the kernel. This
makes such operations more efficient, as half the amount of
KeyInfo is now sent for a
BoundEQ range as before.
(Bug#38793)
17.7.2.14. Changes in MySQL Cluster NDB 6.4.3 (5.1.32-ndb-6.4.3) (23 February 2009)
This is a new Beta development release, incorporating new
features in the NDBCLUSTER storage
engine and fixing recently discovered bugs in MySQL Cluster NDB
6.4.2.
Obtaining MySQL Cluster NDB 6.4.3.
MySQL Cluster NDB 6.4.3 is a source-only release. You can
obtain the source code from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.32-ndb-6.4.3/.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 6.4 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.32 (see
Section C.1.17, “Changes in MySQL 5.1.32 (14 February 2009)”).
Note
This Beta release, as any other pre-production release, should
not be installed on production level
systems or systems with critical data. Please refer to our bug
database at http://bugs.mysql.com/ for more
details about the individual bugs fixed in this version.
Functionality added or changed: Important Change: Replication:
RESET MASTER and
RESET SLAVE now reset the values
shown for Last_IO_Error ,
Last_IO_Errno ,
Last_SQL_Error , and
Last_SQL_Errno in the output of
SHOW SLAVE STATUS .
(Bug#34654) See also Bug#44270.
A new data node configuration parameter
MaxLCPStartDelay has been introduced to
facilitate parallel node recovery by causing a local checkpoint
to be delayed while recovering nodes are synchronizing data
dictionaries and other meta-information. For more information
about this parameter, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
(Bug#43053)
New options are introduced for ndb_restore
for determining which tables or databases should be restored:
--include-tables and
--include-databases can be used to restore
specific tables or databases.
--exclude-tables and
--exclude-databases can be used to exclude
the specified tables or databases from being restored.
For more information about these options, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#40429) Disk Data:
It is now possible to specify default locations for Disk Data
data files and undo log files, either together or separately,
using the data node configuration parameters
FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . For information
about these configuration parameters, see
Disk
Data filesystem parameters.
It is also now possible to specify a log file group, tablespace,
or both, that is created when the cluster is started, using the
InitialLogFileGroup and
InitialTablespace data node configuration
parameters. For information about these configuration
parameters, see
Disk
Data object creation parameters.
Bugs fixed: Performance:
Updates of the SYSTAB_0 system table to
obtain a unique identifier did not use transaction hints for
tables having no primary key. In such cases the NDB kernel used
a cache size of 1. This meant that each insert into a table not
having a primary key required an update of the corresponding
SYSTAB_0 entry, creating a potential
performance bottleneck.
With this fix, inserts on NDB tables without
primary keys can be under some conditions be performed up to
100% faster than previously.
(Bug#39268) Important Note:
It is not possible in this release to install the
InnoDB plugin if
InnoDB support has been compiled
into mysqld.
(Bug#42610) This regression was introduced by Bug#29263. Packaging:
Packages for MySQL Cluster were missing the
libndbclient.so and
libndbclient.a files.
(Bug#42278) Partitioning:
Executing ALTER TABLE ... REORGANIZE
PARTITION on an
NDBCLUSTER table having only one
partition caused mysqld to crash.
(Bug#41945) See also Bug#40389.
Backup IDs greater than 231 were not
handled correctly, causing negative values to be used in backup
directory names and printouts.
(Bug#43042)
When using ndbmtd, NDB kernel threads could
hang while trying to start the data nodes with
LockPagesInMainMemory set to 1.
(Bug#43021)
When using multiple management servers and starting several API
nodes (possibly including one or more SQL nodes) whose
connectstrings listed the management servers in different order,
it was possible for 2 API nodes to be assigned the same node ID.
When this happened it was possible for an API node not to get
fully connected, consequently producing a number of errors whose
cause was not easily recognizable.
(Bug#42973)
When using multi-threaded data nodes,
IndexMemory ,
MaxNoOfLocalOperations , and
MaxNoOfLocalScans were effectively multiplied
by the number of local query handlers in use by each
ndbmtd instance.
(Bug#42765) See also Bug#42215.
ndb_error_reporter worked correctly only with
GNU tar. (With other versions of
tar, it produced empty archives.)
(Bug#42753)
Triggers on NDBCLUSTER tables
caused such tables to become locked.
(Bug#42751) See also Bug#16229, Bug#18135.
When performing more than 32 index or tuple scans on a single
fragment, the scans could be left hanging. This caused
unnecessary timeouts, and in addition could possibly lead to a
hang of an LCP.
(Bug#42559)
A data node failure that occurred between calls to
NdbIndexScanOperation::readTuples(SF_OrderBy)
and NdbTransaction::Execute() was not
correctly handled; a subsequent call to
nextResult() caused a null pointer to be
deferenced, leading to a segfault in mysqld.
(Bug#42545)
If the cluster configuration cache file was larger than 32K, the
management server would not start.
(Bug#42543)
Issuing SHOW GLOBAL STATUS LIKE 'NDB%' before
mysqld had connected to the cluster caused a
segmentation fault.
(Bug#42458)
When using ndbmtd for all data nodes,
repeated failures of one data node during DML operations caused
other data nodes to fail.
(Bug#42450)
Data node failures that occurred before all data nodes had
connected to the cluster were not handled correctly, leading to
additional data node failures.
(Bug#42422)
When using multi-threaded data nodes, their
DataMemory and IndexMemory
usage as reported was multiplied by the number of local query
handlers (worker threads), making it appear that much more
memory was being used than was actually the case.
(Bug#42215) See also Bug#42765.
Given a MySQL Cluster containing no data (that is, whose data
nodes had all been started using --initial , and
into which no data had yet been imported) and having an empty
backup directory, executing START BACKUP with
a user-specified backup ID caused the data nodes to crash.
(Bug#41031)
In some cases, NDB did not check
correctly whether tables had changed before trying to use the
query cache. This could result in a crash of the debug MySQL
server.
(Bug#40464) Disk Data:
It was not possible to add an in-memory column online to a table
that used a table-level or column-level STORAGE
DISK option. The same issue prevented ALTER
ONLINE TABLE ... REORGANIZE PARTITION from working on
Disk Data tables.
(Bug#42549) Disk Data:
Repeated insert and delete operations on disk-based tables could
lead to failures in the NDB Tablespace Manager
(TSMAN kernel block).
(Bug#40344) Disk Data:
Creating a Disk Data tablespace with a very large extent size
caused the data nodes to fail. The issue was observed when using
extent sizes of 100 MB and larger.
(Bug#39096) Disk Data:
Trying to execute a CREATE LOGFILE
GROUP statement using a value greater than
150M for UNDO_BUFFER_SIZE
caused data nodes to crash.
As a result of this fix, the upper limit for
UNDO_BUFFER_SIZE is now
600M ; attempting to set a higher value now
fails gracefully with an error.
(Bug#34102) See also Bug#36702. Disk Data:
When attempting to create a tablespace that already existed, the
error message returned was Table or index with given
name already exists.
(Bug#32662) Disk Data:
Using a path or filename longer than 128 characters for Disk
Data undo log files and tablespace data files caused a number of
issues, including failures of CREATE
LOGFILE GROUP , ALTER LOGFILE
GROUP , CREATE
TABLESPACE , and ALTER
TABLESPACE statements, as well as crashes of
management nodes and data nodes.
With this fix, the maximum length for path and file names used
for Disk Data undo log files and tablespace data files is now
the same as the maximum for the operating system.
(Bug#31769, Bug#31770, Bug#31772) Disk Data:
Attempting to perform a system restart of the cluster where
there existed a logfile group without and undo log files caused
the data nodes to crash.
Note
While issuing a CREATE LOGFILE
GROUP statement without an ADD
UNDOFILE option fails with an error in the MySQL
server, this situation could arise if an SQL node failed
during the execution of a valid CREATE
LOGFILE GROUP statement; it is also possible to
create a logfile group without any undo log files using the
NDB API.
(Bug#17614) Cluster Replication:
Being disconnected from the cluster while setting up the binary
log caused mysqld to hang or crash.
(Bug#43045) Cluster Replication:
Primary key updates on MyISAM and
InnoDB tables failed to replicate
to NDBCLUSTER tables.
(Bug#42921) Cluster API:
Some error messages from ndb_mgmd contained
newline (\n ) characters. This could break the
MGM API protocol, which uses the newline as a line separator.
(Bug#43104) Cluster API:
When using an ordered index scan without putting all key columns
in the read mask, this invalid use of the NDB API went
undetected, which resulted in the use of uninitialized memory.
(Bug#42591)
17.7.2.15. Changes in MySQL Cluster NDB 6.4.2 (5.1.31-ndb-6.4.2) (28 January 2009)
This is a new Beta development release, incorporating new
features in the NDBCLUSTER storage
engine and fixing recently discovered bugs in MySQL Cluster NDB
6.4.1.
Obtaining MySQL Cluster NDB 6.4.2.
MySQL Cluster NDB 6.4.2 is a source-only release. You can
obtain the source code from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.31-ndb-6.4.2/.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 6.4 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.31 (see
Section C.1.19, “Changes in MySQL 5.1.31 (19 January 2009)”).
Note
This Beta release, as any other pre-production release, should
not be installed on production level
systems or systems with critical data. Please refer to our bug
database at http://bugs.mysql.com/ for more
details about the individual bugs fixed in this version.
Bugs fixed:
Connections using IPv6 were not handled correctly by
mysqld.
(Bug#42413) See also Bug#42412, Bug#38247.
When a cluster backup failed with Error 1304 (Node
node_id1 : Backup request from
node_id2 failed to start), no clear
reason for the failure was provided.
As part of this fix, MySQL Cluster now retries backups in the
event of sequence errors.
(Bug#42354) See also Bug#22698.
Issuing SHOW ENGINE
NDBCLUSTER STATUS on an SQL node before the management
server had connected to the cluster caused
mysqld to crash.
(Bug#42264)
When using ndbmtd, setting
MaxNoOfThreads to a value higher than the
actual number of cores available and with insufficient
SharedGlobalMemory caused the data nodes to
crash.
The fix for this issue changes the behavior of
ndbmtd such that its internal job buffers no
longer rely on SharedGlobalMemory .
(Bug#42254)
17.7.2.16. Changes in MySQL Cluster NDB 6.4.1 (5.1.31-ndb-6.4.1) (21 January 2009)
This is a new Beta development release, incorporating new
features in the NDBCLUSTER storage
engine and fixing recently discovered bugs in MySQL Cluster NDB
6.4.0.
Obtaining MySQL Cluster NDB 6.4.1.
MySQL Cluster NDB 6.4.1 is a source-only release. You can
obtain the source code from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.31-ndb-6.4.1/.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, 6.3, and 6.4 releases, as
well as all bugfixes and feature changes which were added in
mainline MySQL 5.1 through MySQL 5.1.31 (see
Section C.1.19, “Changes in MySQL 5.1.31 (19 January 2009)”).
Note
This Beta release, as any other pre-production release, should
not be installed on production level
systems or systems with critical data. Please refer to our bug
database at http://bugs.mysql.com/ for more
details about the individual bugs fixed in this version.
Functionality added or changed: Important Change:
Formerly, when the management server failed to create a
transporter for a data node connection,
net_write_timeout seconds
elapsed before the data node was actually allowed to disconnect.
Now in such cases the disconnection occurs immediately.
(Bug#41965) See also Bug#41713.
Formerly, when using MySQL Cluster Replication, records for
“empty” epochs — that is, epochs in which no
changes to NDBCLUSTER data or
tables took place — were inserted into the
ndb_apply_status and
ndb_binlog_index tables on the slave even
when --log-slave-updates was
disabled. Beginning with MySQL Cluster NDB 6.2.16 and MySQL
Cluster NDB 6.3.13 this was changed so that these
“empty” eopchs were no longer logged. However, it
is now possible to re-enable the older behavior (and cause
“empty” epochs to be logged) by using the
--ndb-log-empty-epochs option. For more
information, see Section 16.1.3.3, “Replication Slave Options and Variables”.
See also Bug#37472. Cluster Replication:
IPv6 networking is now supported between MySQL Cluster SQL
nodes. This means that it is now possible to replicate between
instances of MySQL Cluster using IPv6 addresses.
Important
Currently, other MySQL Cluster processes
(ndbd, ndbmtd,
ndb_mgmd, and ndb_mgm)
do not support IPv6 connections. This means that all MySQL
Cluster data nodes, management servers, and management clients
must connect to and be accessible from one another using IPv4.
In addition, SQL nodes must use IPv4 to communicate with the
cluster. There is also not yet any support in the NDB and MGM
APIs for IPv6, which means that applications written using the
MySQL Cluster APIs must make connections using IPv4. For more
information, see
Section 17.6.3, “Known Issues in MySQL Cluster Replication”.
Bugs fixed:
A maximum of 11 TUP scans were allowed in
parallel.
(Bug#42084)
The management server could hang after attempting to halt it
with the STOP command in the management
client.
(Bug#42056) See also Bug#40922.
When using ndbmtd, one thread could flood
another thread, which would cause the system to stop with a
job buffer full condition (currently
implemented as an abort). This could be caused by committing or
aborting a large transaction (50000 rows or more) on a single
data node running ndbmtd. To prevent this
from happening, the number of signals that can be accepted by
the system threads is calculated before excuting them, and only
executing them if sufficient space is found.
(Bug#42052)
MySQL Cluster would not compile when using
libwrap . This issue was known to occur only
in MySQL Cluster NDB 6.4.0.
(Bug#41918)
Trying to execute an
ALTER ONLINE TABLE
... ADD COLUMN statement while inserting rows into the
table caused mysqld to crash.
(Bug#41905)
When a data node connects to the management server, the node
sends its node ID and transporter type; the management server
then verifies that there is a transporter set up for that node
and that it is in the correct state, and then sends back an
acknowledgement to the connecting node. If the transporter was
not in the correct state, no reply was sent back to the
connecting node, which would then hang until a read timeout
occurred (60 seconds). Now, if the transporter is not in the
correct state, the management server acknowledges this promptly,
and the node immediately disconnects.
(Bug#41713) See also Bug#41965.
Issuing EXIT in the management client
sometimes caused the client to hang.
(Bug#40922)
In the event that a MySQL Cluster backup failed due to file
permissions issues, conflicting reports were issued in the
management client.
(Bug#34526)
If all data nodes were shut down, MySQL clients were unable to
access NDBCLUSTER tables and data
even after the data nodes were restarted, unless the MySQL
clients themselves were restarted.
(Bug#33626)
17.7.2.17. Changes in MySQL Cluster NDB 6.4.0 (5.1.30-ndb-6.4.0) (22 December 2008 Beta)
This is a new Beta development release, incorporating new
features in the NDBCLUSTER storage
engine and fixing recently discovered bugs in previous MySQL
Cluster releases.
Obtaining MySQL Cluster NDB 6.4.0.
MySQL Cluster NDB 6.4.0 is a source-only release. You can
obtain the source code from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.30-ndb-6.4.0/.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1, 6.2, and 6.3 releases, as well
as all bugfixes and feature changes which were added in mainline
MySQL 5.1 through MySQL 5.1.30 (see
Section C.1.20, “Changes in MySQL 5.1.30 (14 November 2008 General Availability)”).
Note
This Beta release, as any other pre-production release, should
not be installed on production level
systems or systems with critical data. Please refer to our bug
database at http://bugs.mysql.com/ for more
details about the individual bugs fixed in this version.
Functionality added or changed: Important Change:
MySQL Cluster now caches its configuration data. This means
that, by default, the management server only reads the global
configuration file (usually named
config.ini ) the first time that it is
started, and does not automatically re-read the this file when
restarted. This behavior can be controlled using new management
server options (--config-dir ,
--initial , and --reload ) that
have been added for this purpose. For more information, see
Section 17.3.2, “MySQL Cluster Configuration Files”, and
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
It is now possible while in Single User Mode to restart all data
nodes using ALL RESTART in the management
client. Restarting of individual nodes while in Single User Mode
remains disallowed.
(Bug#31056)
It is now possible to add data nodes to a MySQL Cluster online
— that is, to a running MySQL Cluster without shutting it
down.
For information about the procedure for adding data nodes
online, see Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
A multi-threaded version of the MySQL Cluster data node daemon
is now available. The multi-threaded ndbmtd
binary is similar to ndbd and functions in
much the same way, but is intended for use on machines with
multiple CPU cores.
For more information, see
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”.
It is now possible when performing a cluster backup to determine
whether the backup matches the state of the data when the backup
began or when it ended, using the new START
BACKUP options SNAPSHOTSTART and
SNAPSHOTEND in the management client. See
Section 17.5.3.2, “Using The MySQL Cluster Management Client to Create a Backup”,
for more information.
Bugs fixed:
API nodes disconnected too agressively from cluster when data
nodes were being restarted. This could sometimes lead to the API
node being unable to access the cluster at all during a rolling
restart.
(Bug#41462)
When long signal buffer exhaustion in the
ndbd process resulted in a signal being
dropped, the usual handling mechanism did not take fragmented
signals into account. This could result in a crash of the data
node because the fragmented signal handling mechanism was not
able to work with the missing fragments.
(Bug#39235)
The failure of a master node during a DDL operation caused the
cluster to be unavailable for further DDL operations until it
was restarted; failures of nonmaster nodes during DLL operations
caused the cluster to become completely inaccessible.
(Bug#36718)
Status messages shown in the management client when restarting a
management node were inappropriate and misleading. Now, when
restarting a management node, the messages displayed are as
follows, where node_id is the
management node's node ID:
ndb_mgm> node_id RESTART
Shutting down MGM node node_id for restart
Node node_id is being restarted
ndb_mgm>
(Bug#29275)
A data node failure when NoOfReplicas was
greater than 2 caused all cluster SQL nodes to crash.
(Bug#18621) Partitioning:
A query on a user-partitioned table caused MySQL to crash, where
the query had the following characteristics:
The query's WHERE clause referenced
an indexed column that was also in the partitioning key.
The query's WHERE clause included a
value found in the partition.
The query's WHERE clause used the
< or <>
operators to compare with the indexed column's value
with a constant.
The query used an ORDER BY clause, and
the same indexed column was used in the ORDER
BY clause.
The ORDER BY clause used an explcit or
implicit ASC sort priority.
Two examples of such a query are given here, where
a represents an indexed column used in the
table's partitioning key:
SELECT * FROM table WHERE a < constant ORDER BY a;
SELECT * FROM table WHERE a <> constant ORDER BY a;
(Bug#40954) This regression was introduced by Bug#30573, Bug#33257, Bug#33555.
17.7.3. Changes in MySQL Cluster NDB 6.3
This section contains change history information for MySQL Cluster
releases based on version 6.3 of the
NDBCLUSTER storage engine.
For an overview of new features added in MySQL Cluster NDB 6.3,
see Section 17.1.4.4, “MySQL Cluster Development in MySQL Cluster NDB 6.3”.
17.7.3.1. Changes in MySQL Cluster NDB 6.3.32 (5.1.41-ndb-6.3.32) (Not yet released)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.41 (see Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
When using NoOfReplicas equal to 1 or 2, if
data nodes from one node group were restarted 256 times and
applications were running traffic such that it would encounter
NDB error 1204
(Temporary failure, distribution
changed), the live node in the node group would
crash, causing the cluster to crash as well. The crash occurred
only when the error was encountered on the 256th restart; having
the error on any previous or subsequent restart did not cause
any problems.
(Bug#50930)
17.7.3.2. Changes in MySQL Cluster NDB 6.3.31 (5.1.41-ndb-6.3.31) (02 February 2010)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.41 (see Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Change:
The maximum allowed value of the
ndb_autoincrement_prefetch_sz
system variable has been increased from 256 to 65536.
(Bug#50621) Cluster Replication:
Due to the fact that no timestamp is available for delete
operations, a delete using NDB$MAX() is
actually processed as NDB$OLD . However,
because this is not optimal for some use cases,
NDB$MAX_DELETE_WIN() is added as a conflict
resolution function; if the “timestamp” column
value for a given row adding or updating an existing row coming
from the master is higher than that on the slave, it is applied
(as with NDB$MAX() ); however, delete
operations are treated as always having the higher value.
See
NDB$MAX_DELETE_WIN(column_name ),
for more information.
(Bug#50650)
Bugs fixed:
Setting BuildIndexThreads greater than 1 with
more than 31 ordered indexes caused node and system restarts to
fail.
(Bug#50266)
Dropping unique indexes in parallel while they were in use could
cause node and cluster failures.
(Bug#50118)
When setting the LockPagesInMainMemory
configuration parameter failed, only the error Failed
to memlock pages... was returned. Now in such cases
the operating system's error code is also returned.
(Bug#49724)
If a query on an NDB table compared
a constant string value to a column, and the length of the
string was greater than that of the column, condition pushdown
did not work correctly. (The string was truncated to fit the
column length before being pushed down.) Now in such cases, the
condition is no longer pushed down.
(Bug#49459)
Performing intensive inserts and deletes in parallel with a high
scan load could a data node crashes due to a failure in the
DBACC kernel block. This was because checking
for when to perform bucket splits or merges considered the first
4 scans only.
(Bug#48700)
During Start Phases 1 and 2, the STATUS
command sometimes (falsely) returned Not
Connected for data nodes running
ndbmtd.
(Bug#47818)
When performing a DELETE that
included a left join from an NDB
table, only the first matching row was deleted.
(Bug#47054)
When setting LockPagesInMainMemory , the
stated memory was not allocated when the node was started, but
rather only when the memory was used by the data node process
for other reasons.
(Bug#37430)
Trying to insert more rows than would fit into an
NDB table caused data nodes to crash. Now in
such situations, the insert fails gracefully with error 633
Table fragment hash index has reached maximum
possible size.
(Bug#34348) Disk Data:
When a crash occurs due to a problem in Disk Data code, the
currently active page list is printed to
stdout (that is, in one or more
ndb_nodeid _out.log
files). One of these lists could contain an endless loop; this
caused a printout that was effectively never-ending. Now in such
cases, a maximum of 512 entries is printed from each list.
(Bug#42431)
On Mac OS X or Windows, sending a SIGHUP
signal to the server or an asynchronous flush (triggered by
flush_time ) caused the server
to crash.
(Bug#47525)
The ARCHIVE storage engine lost
records during a bulk insert.
(Bug#46961)
When using the ARCHIVE storage
engine, SHOW TABLE STATUS displayed incorrect
information for Max_data_length ,
Data_length and
Avg_row_length .
(Bug#29203)
17.7.3.3. Changes in MySQL Cluster NDB 6.3.30 (5.1.39-ndb-6.3.30) (15 December 2009)
This special-purpose evaluation release incorporates all
bugfixes and feature changes from previous MySQL Cluster NDB 6.3
releases plus one new feature for evaluation. Most users of
MySQL Cluster NDB 6.3 do not require this release; instead, you
should use MySQL Cluster NDB 6.3.29 until MySQL Cluster NDB
6.3.31 is released.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.30 from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.39-ndb-6.3.30/.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.39 (see Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: 17.7.3.4. Changes in MySQL Cluster NDB 6.3.29 (5.1.39-ndb-6.3.29) (15 December 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.29 from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.39-ndb-6.3.29/.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.39 (see Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
This enhanced functionality is supported for upgrades to MySQL
Cluster NDB 7.0 when the NDB engine
version is 7.0.10 or later.
(Bug#48528, Bug#49163)
The output from ndb_config
--configinfo
--xml now indicates, for each
configuration parameter, the following restart type information:
(Bug#47366)
Bugs fixed:
Node takeover during a system restart occurs when the REDO log
for one or more data nodes is out of date, so that a node
restart is invoked for that node or those nodes. If this happens
while a mysqld process is attached to the
cluster as an SQL node, the mysqld takes a
global schema lock (a row lock), while trying to set up
cluster-internal replication.
However, this setup process could fail, causing the global
schema lock to be held for an excessive length of time, which
made the node restart hang as well. As a result, the mysqld
failed to set up cluster-internal replication, which led to
tables being read-only, and caused one node to hang during the
restart.
Note
This issue could actually occur in MySQL Cluster NDB 7.0 only,
but the fix was also applied MySQL Cluster NDB 6.3, in order
to keep the two codebases in alignment.
(Bug#49560)
Sending SIGHUP to a mysqld
running with the --ndbcluster and
--log-bin options caused the
process to crash instead of refreshing its log files.
(Bug#49515)
If the master data node receiving a request from a newly-started
API or data node for a node ID died before the request has been
handled, the management server waited (and kept a mutex) until
all handling of this node failure was complete before responding
to any other connections, instead of responding to other
connections as soon as it was informed of the node failure (that
is, it waited until it had received a NF_COMPLETEREP signal
rather than a NODE_FAILREP signal). On visible effect of this
misbehavior was that it caused management client commands such
as SHOW and ALL STATUS to respond with unnecessary slowness in
such circumstances.
(Bug#49207)
When evaluating the options
--include-databases ,
--include-tables ,
--exclude-databases , and
--exclude-tables , the
ndb_restore program overwrote the result of
the database-level options with the result of the table-level
options rather than merging these results together, sometimes
leading to unexpected and unpredictable results.
As part of the fix for this problem, the semantics of these
options have been clarified; because of this, the rules
governing their evaluation have changed slightly. These changes
be summed up as follows:
All --include-* and
--exclude-* options are now evaluated from
right to left in the order in which they are passed to
ndb_restore.
All --include-* and
--exclude-* options are now cumulative.
In the event of a conflict, the first (rightmost) option
takes precedence.
For more detailed information and examples, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#48907)
When performing tasks that generated large amounts of I/O (such
as when using ndb_restore), an internal
memory buffer could overflow, causing data nodes to fail with
signal 6.
Subsequent analysis showed that this buffer was not actually
required, so this fix removes it.
(Bug#48861)
Exhaustion of send buffer memory or long signal memory caused
data nodes to crash. Now an appropriate error message is
provided instead when this situation occurs.
(Bug#48852)
Under certain conditions, accounting of the number of free scan
records in the local query handler could be incorrect, so that
during node recovery or a local checkpoint operations, the LQH
could find itself lacking a scan record that is expected to
find, causing the node to crash.
(Bug#48697) See also Bug#48564.
The creation of an ordered index on a table undergoing DDL
operations could cause a data node crash under certain
timing-dependent conditions.
(Bug#48604)
During an LCP master takeover, when the newly elected master did
not receive a COPY_GCI LCP protocol message
but other nodes participating in the local checkpoint had
received one, the new master could use an uninitialized
variable, which caused it to crash.
(Bug#48584)
When running many parallel scans, a local checkpoint (which
performs a scan internally) could find itself not getting a scan
record, which led to a data node crash. Now an extra scan record
is reserved for this purpose, and a problem with obtaining the
scan record returns an appropriate error (error code 489,
Too many active scans).
(Bug#48564)
During a node restart, logging was enabled on a per-fragment
basis as the copying of each fragment was completed but local
checkpoints were not enabled until all fragments were copied,
making it possible to run out of redo log file space
(NDB error code 410) before the
restart was complete. Now logging is enabled only after all
fragments has been copied, just prior to enabling local
checkpoints.
(Bug#48474)
When employing NDB native backup to
back up and restore an empty NDB
table that used a non-sequential
AUTO_INCREMENT value, the
AUTO_INCREMENT value was not restored
correctly.
(Bug#48005)
ndb_config --xml
--configinfo now indicates that
parameters belonging in the [SCI] ,
[SCI DEFAULT] , [SHM] , and
[SHM DEFAULT] sections of the
config.ini file are deprecated or
experimental, as appropriate.
(Bug#47365)
NDB stores blob column data in a
separate, hidden table that is not accessible from MySQL. If
this table was missing for some reason (such as accidental
deletion of the file corresponding to the hidden table) when
making a MySQL Cluster native backup, ndb_restore crashed when
attempting to restore the backup. Now in such cases, ndb_restore
fails with the error message Table
table_name has blob column
(column_name ) with missing parts
table in backup instead.
(Bug#47289)
DROP DATABASE failed when there
were stale temporary NDB tables in
the database. This situation could occur if
mysqld crashed during execution of a
DROP TABLE statement after the
table definition had been removed from
NDBCLUSTER but before the
corresponding .ndb file had been removed
from the crashed SQL node's data directory. Now, when
mysqld executes DROP
DATABASE , it checks for these files and removes them
if there are no corresponding table definitions for them found
in NDBCLUSTER .
(Bug#44529)
Creating an NDB table with an
excessive number of large BIT
columns caused the cluster to fail. Now, an attempt to create
such a table is rejected with error 791 (Too many
total bits in bitfields).
(Bug#42046) See also Bug#42047.
When a long-running transaction lasting long enough to cause
Error 410 (REDO log files overloaded) was
later committed or rolled back, it could happen that
NDBCLUSTER was not able to release
the space used for the REDO log, so that the error condition
persisted indefinitely.
The most likely cause of such transactions is a bug in the
application using MySQL Cluster. This fix should handle most
cases where this might occur.
(Bug#36500)
Deprecation and usage information obtained from
ndb_config --configinfo
regarding the PortNumber and
ServerPort configuration parameters was
improved.
(Bug#24584) Disk Data:
When running a write-intensive workload with a very large disk
page buffer cache, CPU usage approached 100% during a local
checkpoint of a cluster containing Disk Data tables.
(Bug#49532) Disk Data:
Repeatedly creating and then dropping Disk Data tables could
eventually lead to data node failures.
(Bug#45794, Bug#48910) Disk Data:
When the FileSystemPathUndoFiles
configuration parameter was set to an non-existent path, the
data nodes shut down with the generic error code 2341
(Internal program error). Now in such
cases, the error reported is error 2815 (File not
found).
Cluster Replication:
When expire_logs_days was set,
the thread performing the purge of the log files could deadlock,
causing all binary log operations to stop.
(Bug#49536) Cluster API:
When a DML operation failed due to a uniqueness violation on an
NDB table having more than one
unique index, it was difficult to determine which constraint
caused the failure; it was necessary to obtain an
NdbError object, then decode its
details property, which in could lead to
memory management issues in application code.
To help solve this problem, a new API method
Ndb::getNdbErrorDetail() is added, providing
a well-formatted string containing more precise information
about the index that caused the unque constraint violation. The
following additional changes are also made in the NDB API:
For more information, see
Ndb::getNdbErrorDetail() ,
The NdbError Structure, and
Dictionary::listObjects() .
(Bug#48851) Cluster API:
When using blobs, calling getBlobHandle()
requires the full key to have been set using
equal() , because
getBlobHandle() must access the key for
adding blob table operations. However, if
getBlobHandle() was called without first
setting all parts of the primary key, the application using it
crashed. Now, an appropriate error code is returned instead.
(Bug#28116, Bug#48973)
17.7.3.5. Changes in MySQL Cluster NDB 6.3.28b (5.1.39-ndb-6.3.28b) (10 November 2009)
This release includes a fix for Bug#48651, which was discovered
in MySQL Cluster NDB 6.3.28a shortly after release. The MySQL
Cluster NDB 6.3.28b release is identical in all other respects
to MySQL Cluster NDB 6.3.28a. Users who have already installed
MySQL Cluster NDB 6.3.28a should upgrade to MySQL Cluster NDB
6.3.28b as soon as possible; users seeking to upgrade from any
other previous MySQL Cluster 6.3 release should upgrade to MySQL
Cluster NDB 6.3.28b instead.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.28b from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.39 (see Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.3.6. Changes in MySQL Cluster NDB 6.3.28a (5.1.39-ndb-6.3.28a) (05 November 2009)Important
MySQL Cluster NDB 6.3.28a was pulled shortly after release due
to Bug#48651. Users seeking to upgrade from a previous MySQL
Cluster NDB 6.3 release should instead use MySQL Cluster NDB
6.3.28b, which contains a fix for this bug, in addition to all
bugfixes and improvements made in MySQL Cluster NDB 6.3.28a.
This release includes a fix for Bug#48531, which was discovered
in MySQL Cluster NDB 6.3.28 shortly after release. The MySQL
Cluster NDB 6.3.28a release is identical in all other respects
to MySQL Cluster NDB 6.3.28. Users who have already installed
MySQL Cluster NDB 6.3.28 should upgrade to MySQL Cluster NDB
6.3.28a as soon as possible; users seeking to upgrade from any
other previous MySQL Cluster 6.3 release should upgrade to MySQL
Cluster NDB 6.3.28a instead.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.39 (see Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.3.7. Changes in MySQL Cluster NDB 6.3.28 (5.1.39-ndb-6.3.28) (31 October 2009)Important
MySQL Cluster NDB 6.3.28 and 6.3.28a were pulled shortly after
being released due to Bug#48531 and Bug#48651. Users seeking
to upgrade from a previous MySQL Cluster NDB 6.3 release
should instead use MySQL Cluster NDB 6.3.28b, which contains
fixes for these critical bugs, in addition to all bugfixes and
improvements made in MySQL Cluster NDB 6.3.28.
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.39 (see Section C.1.8, “Changes in MySQL 5.1.39 (04 September 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Performance:
Significant improvements in redo log handling and other
filesystem operations can yield a considerable reduction in the
time required for restarts. While actual restart times observed
in a production setting will naturally vary according to
database size, hardware, and other conditions, our own
preliminary testing shows that these optimizations can yield
startup times that are faster than those typical of previous
MySQL Cluster releases by a factor of 50 or more.
Bugs fixed: Important Change:
The --with-ndb-port-base option for
configure did not function correctly, and has
been deprecated. Attempting to use this option produces the
warning Ignoring deprecated option
--with-ndb-port-base.
Beginning with MySQL Cluster NDB 7.1.0, the deprecation warning
itself is removed, and the --with-ndb-port-base
option is simply handled as an unknown and invalid option if you
try to use it.
(Bug#47941) See also Bug#38502.
In certain cases, performing very large inserts on
NDB tables when using
ndbmtd caused the memory allocations for
ordered or unique indexes (or both) to be exceeded. This could
cause aborted transactions and possibly lead to data node
failures.
(Bug#48037) See also Bug#48113.
For UPDATE
IGNORE statements, batching of updates is now
disabled. This is because such statements failed when batching
of updates was employed if any updates violated a unique
constraint, to the fact a unique constraint violation could not
be handled without aborting the transaction.
(Bug#48036)
Starting a data node with a very large amount of
DataMemory (approximately 90G or more) could
lead to crash of the node due to job buffer congestion.
(Bug#47984)
When an UPDATE statement was
issued against an NDB table where
an index was used to identify rows but no data was actually
changed, the NDB storage returned
zero found rows.
For example, consider the table created and populated using
these statements:
CREATE TABLE t1
(
c1 INT NOT NULL,
c2 INT NOT NULL,
PRIMARY KEY(c1),
KEY(c2)
)
ENGINE = NDB;
INSERT INTO t1 VALUES(1, 1);
The following UPDATE statements,
even though they did not change any rows, each still matched a
row, but this was reported incorrectly in both cases, as shown
here:
mysql> UPDATE t1 SET c2 = 1 WHERE c1 = 1;
Query OK, 0 rows affected (0.00 sec)
Rows matched: 0 Changed: 0 Warnings: 0
mysql> UPDATE t1 SET c1 = 1 WHERE c2 = 1;
Query OK, 0 rows affected (0.00 sec)
Rows matched: 0 Changed: 0 Warnings: 0
Now in such cases, the number of rows matched is correct. (In
the case of each of the example
UPDATE statements just shown,
this is displayed as Rows matched: 1, as it should be.)
This issue could affect UPDATE
statements involving any indexed columns in
NDB tables, regardless of the type
of index (including KEY , UNIQUE
KEY , and PRIMARY KEY ) or the number
of columns covered by the index.
(Bug#47955)
On Solaris, shutting down a management node failed when issuing
the command to do so from a client connected to a different
management node.
(Bug#47948)
Setting FragmentLogFileSize to a value
greater than 256 MB led to errors when trying to read the redo
log file.
(Bug#47908)
SHOW CREATE TABLE did not display
the AUTO_INCREMENT value for
NDB tables having
AUTO_INCREMENT columns.
(Bug#47865)
Under some circumstances, when a scan encountered an error early
in processing by the DBTC kernel block (see
The DBTC Block), a node
could crash as a result. Such errors could be caused by
applications sending incorrect data, or, more rarely, by a
DROP TABLE operation executed in
parallel with a scan.
(Bug#47831)
When starting a node and synchronizing tables, memory pages were
allocated even for empty fragments. In certain situations, this
could lead to insufficient memory.
(Bug#47782)
A very small race-condition between
NODE_FAILREP and
LQH_TRANSREQ signals when handling node
failure could lead to operations (locks) not being taken over
when they should have been, and subsequently becoming stale.
This could lead to node restart failures, and applications
getting into endless lock-conflicts with operations that were
not released until the node was restarted.
(Bug#47715) See also Bug#41297.
configure failed to honor the
--with-zlib-dir option when trying to build
MySQL Cluster from source.
(Bug#47223)
ndbd was not built correctly when compiled
using gcc 4.4.0. (The ndbd
binary was built, but could not be started.)
(Bug#46113)
If a node failed while sending a fragmented long signal, the
receiving node did not free long signal assembly resources that
it had allocated for the fragments of the long signal that had
already been received.
(Bug#44607)
When starting a cluster with a great many tables, it was
possible for MySQL client connections as well as the slave SQL
thread to issue DML statements against MySQL Cluster tables
before mysqld had finished connecting to the
cluster and making all tables writeable. This resulted in
Table ... is read only errors for clients
and the Slave SQL thread.
This issue is fixed by introducing the
--ndb-wait-setup option for the
MySQL server. This provides a configurable maximum amount of
time that mysqld waits for all
NDB tables to become writeable,
before allowing MySQL clients or the slave SQL thread to
connect.
(Bug#40679) See also Bug#46955.
When building MySQL Cluster, it was possible to configure the
build using --with-ndb-port without supplying a
port number. Now in such cases, configure
fails with an error.
(Bug#38502) See also Bug#47941.
When the MySQL server SQL mode included
STRICT_TRANS_TABLES , storage
engine warnings and error codes specific to
NDB were returned when errors occurred,
instead of the MySQL server errors and error codes expected by
some programming APIs (such as Connector/J) and applications.
(Bug#35990)
When a copying operation exhausted the available space on a data
node while copying large BLOB
columns, this could lead to failure of the data node and a
Table is full error on the SQL node which
was executing the operation. Examples of such operations could
include an ALTER TABLE that
changed an INT column to a
BLOB column, or a bulk insert of
BLOB data that failed due to
running out of space or to a duplicate key error.
(Bug#34583, Bug#48040) See also Bug#41674, Bug#45768. Replication:
When mysqlbinlog
--verbose was used to read a
binary log that had been recorded using the row-based format,
the output for events that updated some but not all columns of
tables was not correct.
(Bug#47323) Disk Data:
A local checkpoint of an empty fragment could cause a crash
during a system restart which was based on that LCP.
(Bug#47832) See also Bug#41915. Cluster Replication:
When using multiple active replication channels, it was
sometimes possible that a node group would fail on the slave
cluster, causing the slave cluster to shut down.
(Bug#47935) Cluster Replication:
When recording a binary log using the
--ndb-log-update-as-write and
--ndb-log-updated-only options
(both enabled by default) and later attempting to apply that
binary log with mysqlbinlog, any operations
that were played back from the log but which updated only some
(but not all) columns caused any columns that were not updated
to be reset to their default values.
(Bug#47674) See also Bug#47323, Bug#46662. Cluster Replication:
mysqlbinlog failed to apply correctly a
binary log that had been recorded using
--ndb-log-update-as-write=1 .
(Bug#46662) See also Bug#47323, Bug#47674. Cluster API:
If an NDB API program reads the same column more than once, it
is possible exceed the maximum allowable message size, in which
case the operation should be aborted due to NDB error 880
Tried to read too much - too many getValue
calls, however due to a change introduced in MySQL
Cluster NDB 6.3.18, the check for this was not done correctly,
which instead caused a data node crash.
(Bug#48266) Cluster API:
The NDB API methods Dictionary::listEvents() ,
Dictionary::listIndexes() ,
Dictionary::listObjects() , and
NdbOperation::getErrorLine() formerly had
both const and non-const
variants. The non-const versions of these
methods have been removed. In addition, the
NdbOperation::getBlobHandle() method has been
re-implemented in order to provide consistent internal
semantics.
(Bug#47798) Cluster API:
A duplicate read of a column caused NDB API applications to
crash.
(Bug#45282) Cluster API:
The error handling shown in the example file
ndbapi_scan.cpp included with the MySQL
Cluster distribution was incorrect.
(Bug#39573)
Installation of MySQL on Windows would fail to set the correct
location for the character set files, which could lead to
mysqld and mysql failing
to initialize properly.
(Bug#17270)
17.7.3.8. Changes in MySQL Cluster NDB 6.3.27a (5.1.37-ndb-6.3.27a) (07 October 2009)
This release includes a fix for Bug#47844, which was discovered
in MySQL Cluster NDB 6.3.27 shortly after release. The MySQL
Cluster NDB 6.3.27a release is identical in all other respects
to MySQL Cluster NDB 6.3.27. Users who have already installed
MySQL Cluster NDB 6.3.27 should upgrade to MySQL Cluster NDB
6.3.27a as soon as possible; users seeking to upgrade from MySQL
Cluster NDB 6.3.26 or another previous MySQL Cluster 6.3 release
should upgrade to MySQL Cluster NDB 6.3.27a instead.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.27a from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.37-ndb-6.3.27a/.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.37 (see Section C.1.11, “Changes in MySQL 5.1.37 (13 July 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
The disconnection of an API or SQL node having a node ID greater
than 49 caused a forced shutdown of the cluster.
(Bug#47844)
The error message text for NDB
error code 410 (REDO log files
overloaded...) was truncated.
(Bug#23662)
17.7.3.9. Changes in MySQL Cluster NDB 6.3.27 (5.1.37-ndb-6.3.27) (30 September 2009)Important
MySQL Cluster NDB 6.3.27 was pulled shortly after release due
to Bug#47844. Users seeking to upgrade from a previous MySQL
Cluster NDB 6.3 release should instead use MySQL Cluster NDB
6.3.27a, which contains a fix for this bug, in addition to all
bugfixes and improvements made in MySQL Cluster NDB 6.3.27.
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.37 (see Section C.1.11, “Changes in MySQL 5.1.37 (13 July 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: Cluster Replication: Important Change:
In a MySQL Cluster acting as a replication slave and having
multiple SQL nodes, only the SQL node receiving events directly
from the master recorded DDL statements in its binary logs
unless this SQL node also had binary logging enabled; otherwise,
other SQL nodes in the slave cluster failed to log DDL
statements, regardless of their individual
--log-bin settings.
The fix for this issue aligns binary logging of DDL statements
with that of DML statements. In particular, you should take note
of the following:
DDL and DML statements on the master cluster are logged with
the server ID of the server that actually writes the log.
DDL and DML statements on the master cluster are logged by
any attached mysqld that has binary
logging enabled.
Replicated DDL and DML statements on the slave are logged by
any attached mysqld that has both
--log-bin and
--log-slave-updates enabled.
Replicated DDL and DML statements are logged with the server
ID of the original (master) MySQL server by any attached
mysqld that has both
--log-bin and
--log-slave-updates enabled.
Affect on upgrades.
When upgrading from a previous MySQL CLuster release, you
should perform either one of the following:
Upgrade servers that are performing binary logging
before those that are not; do not perform any DDL on
“old” SQL nodes until all SQL nodes have
been upgraded.
Make sure that
--log-slave-updates is
enabled on all SQL nodes performing binary logging prior
to the upgrade, so that all DDL is captured.
Note
Logging of DML statements was not affected by this issue.
(Bug#45756)
mysqld allocated an excessively large buffer
for handling BLOB values due to
overestimating their size. (For each row, enough space was
allocated to accommodate every
BLOB or
TEXT column value in the result
set.) This could adversely affect performance when using tables
containing BLOB or
TEXT columns; in a few extreme
cases, this issue could also cause the host system to run out of
memory unexpectedly.
(Bug#47574) See also Bug#47572, Bug#47573.
NDBCLUSTER uses a dynamically-allocated
buffer to store BLOB or
TEXT column data that is read
from rows in MySQL Cluster tables.
When an instance of the NDBCLUSTER table
handler was recycled (this can happen due to table definition
cache pressure or to operations such as
FLUSH TABLES or
ALTER TABLE ), if the last row
read contained blobs of zero length, the buffer was not freed,
even though the reference to it was lost. This resulted in a
memory leak.
For example, consider the table defined and populated as shown
here:
CREATE TABLE t (a INT PRIMARY KEY, b LONGTEXT) ENGINE=NDB;
INSERT INTO t VALUES (1, REPEAT('F', 20000));
INSERT INTO t VALUES (2, '');
Now execute repeatedly a SELECT
on this table, such that the zero-length
LONGTEXT row is
last, followed by a FLUSH
TABLES statement (which forces the handler object to
be re-used), as shown here:
SELECT a, length(b) FROM bl ORDER BY a;
FLUSH TABLES;
Prior to the fix, this resulted in a memory leak proportional to
the size of the stored
LONGTEXT value
each time these two statements were executed.
(Bug#47573) See also Bug#47572, Bug#47574.
Large transactions involving joins between tables containing
BLOB columns used excessive
memory.
(Bug#47572) See also Bug#47573, Bug#47574.
A variable was left uninitialized while a data node copied data
from its peers as part of its startup routine; if the starting
node died during this phase, this could lead a crash of the
cluster when the node was later restarted.
(Bug#47505)
When a data node restarts, it first runs the redo log until
reaching the latest restorable global checkpoint; after this it
scans the remainder of the redo log file, searching for entries
that should be invalidated so they are not used in any
subsequent restarts. (It is possible, for example, if restoring
GCI number 25, that there might be entries belonging to GCI 26
in the redo log.) However, under certain rare conditions, during
the invalidation process, the redo log files themselves were not
always closed while scanning ahead in the redo log. In rare
cases, this could lead to MaxNoOfOpenFiles
being exceeded, causing a the data node to crash.
(Bug#47171)
For very large values of MaxNoOfTables +
MaxNoOfAttributes , the calculation for
StringMemory could overflow when creating
large numbers of tables, leading to NDB error 773
(Out of string memory, please modify StringMemory
config parameter), even when
StringMemory was set to
100 (100 percent).
(Bug#47170)
The default value for the StringMemory
configuration parameter, unlike other MySQL Cluster
configuration parameters, was not set in
ndb/src/mgmsrv/ConfigInfo.cpp .
(Bug#47166)
Signals from a failed API node could be received after an
API_FAILREQ signal (see
Operations and Signals)
has been received from that node, which could result in invalid
states for processing subsequent signals. Now, all pending
signals from a failing API node are processed before any
API_FAILREQ signal is received.
(Bug#47039) See also Bug#44607.
Using triggers on NDB tables caused
ndb_autoincrement_prefetch_sz
to be treated as having the NDB kernel's internal default
value (32) and the value for this variable as set on the
cluster's SQL nodes to be ignored.
(Bug#46712)
Running an ALTER TABLE statement
while an NDB backup was in progress caused
mysqld to crash.
(Bug#44695)
When performing auto-discovery of tables on individual SQL
nodes, NDBCLUSTER attempted to overwrite
existing MyISAM .frm
files and corrupted them.
Workaround.
In the mysql client, create a new table
(t2 ) with same definition as the corrupted
table (t1 ). Use your system shell or file
manager to rename the old .MYD file to
the new file name (for example, mv t1.MYD
t2.MYD). In the mysql client,
repair the new table, drop the old one, and rename the new
table using the old file name (for example,
RENAME TABLE t2
TO t1 ).
(Bug#42614)
Running ndb_restore with the
--print or --print_log option
could cause it to crash.
(Bug#40428, Bug#33040)
An insert on an NDB table was not
always flushed properly before performing a scan. One way in
which this issue could manifest was that
LAST_INSERT_ID() sometimes failed
to return correct values when using a trigger on an
NDB table.
(Bug#38034)
When a data node received a TAKE_OVERTCCONF
signal from the master before that node had received a
NODE_FAILREP , a race condition could in
theory result.
(Bug#37688) See also Bug#25364, Bug#28717.
Some joins on large NDB tables
having TEXT or
BLOB columns could cause
mysqld processes to leak memory. The joins
did not need to reference the
TEXT or
BLOB columns directly for this
issue to occur.
(Bug#36701)
On Mac OS X 10.5, commands entered in the management client
failed and sometimes caused the client to hang, although
management client commands invoked using the
--execute (or
-e ) option from the system shell worked
normally.
For example, the following command failed with an error and hung
until killed manually, as shown here:
ndb_mgm> SHOW
Warning, event thread startup failed, degraded printouts as result, errno=36
^C
However, the same management client command, invoked from the
system shell as shown here, worked correctly:
shell> ndb_mgm -e "SHOW"
(Bug#35751) See also Bug#34438. Replication:
In some cases, a STOP SLAVE
statement could cause the replication slave to crash. This issue
was specific to MySQL on Windows or Macintosh platforms.
(Bug#45238, Bug#45242, Bug#45243, Bug#46013, Bug#46014, Bug#46030) See also Bug#40796. Disk Data:
Calculation of free space for Disk Data table fragments was
sometimes done incorrectly. This could lead to unnecessary
allocation of new extents even when sufficient space was
available in existing ones for inserted data. In some cases,
this might also lead to crashes when restarting data nodes.
Note
This miscalculation was not reflected in the contents of the
INFORMATION_SCHEMA.FILES table,
as it applied to extents allocated to a fragment, and not to a
file.
(Bug#47072) Cluster API:
In some circumstances, if an API node encountered a data node
failure between the creation of a transaction and the start of a
scan using that transaction, then any subsequent calls to
startTransaction() and
closeTransaction() could cause the same
transaction to be started and closed repeatedly.
(Bug#47329) Cluster API:
Performing multiple operations using the same primary key within
the same
NdbTransaction::execute()
call could lead to a data node crash.
Note
This fix does not make change the fact that performing
multiple operations using the same primary key within the same
execute() is not supported; because there
is no way to determine the order of such operations, the
result of such combined operations remains undefined.
(Bug#44065) See also Bug#44015. API:
The fix for Bug#24507 could lead in some cases to client
application failures due to a race condition. Now the server
waits for the “dummy” thread to return before
exiting, thus making sure that only one thread can initialize
the POSIX threads library.
(Bug#42850)
17.7.3.10. Changes in MySQL Cluster NDB 6.3.26 (5.1.35-ndb-6.3.26) (26 August 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.35 (see Section C.1.13, “Changes in MySQL 5.1.35 (13 May 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
On Solaris platforms, the MySQL Cluster management server and
NDB API applications now use CLOCK_REALTIME
as the default clock.
(Bug#46183)
A new option --exclude-missing-columns has been
added for the ndb_restore program. In the
event that any tables in the database or databases being
restored to have fewer columns than the same-named tables in the
backup, the extra columns in the backup's version of the
tables are ignored. For more information, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#43139) Note
This issue, originally resolved in MySQL 5.1.16, re-occurred
due to a later (unrelated) change. The fix has been
re-applied.
(Bug#25984)
Bugs fixed:
Restarting the cluster following a local checkpoint and an
online ALTER TABLE on a non-empty
table caused data nodes to crash.
(Bug#46651)
Full table scans failed to execute when the cluster contained
more than 21 table fragments.
Note
The number of table fragments in the cluster can be calculated
as the number of data nodes, times 8 (that is, times the value
of the internal constant
MAX_FRAG_PER_NODE ), divided by the number
of replicas. Thus, when NoOfReplicas = 1 at
least 3 data nodes were required to trigger this issue, and
when NoOfReplicas = 2 at least 4 data nodes
were required to do so.
(Bug#46490)
Killing MySQL Cluster nodes immediately following a local
checkpoint could lead to a crash of the cluster when later
attempting to perform a system restart.
The exact sequence of events causing this issue was as follows:
Local checkpoint occurs.
Immediately following the LCP, kill the master data node.
Kill the remaining data nodes within a few seconds of
killing the master.
Attempt to restart the cluster.
(Bug#46412)
Ending a line in the config.ini file with
an extra semicolon character (; ) caused
reading the file to fail with a parsing error.
(Bug#46242)
When combining an index scan and a delete with a primary key
delete, the index scan and delete failed to initialize a flag
properly. This could in rare circumstances cause a data node to
crash.
(Bug#46069)
OPTIMIZE TABLE on an
NDB table could in some cases cause
SQL and data nodes to crash. This issue was observed with both
ndbd and ndbmtd.
(Bug#45971)
The AutoReconnect configuration parameter for
API nodes (including SQL nodes) has been added. This is intended
to prevent API nodes from re-using allocated node IDs during
cluster restarts. For more information, see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
This fix also introduces two new methods of the
Ndb_cluster_connection class in the NDB API.
For more information, see
Ndb_cluster_connection::set_auto_reconnect() ,
and
Ndb_cluster_connection::get_auto_reconnect() .
(Bug#45921)
The signals used by ndb_restore to send
progress information about backups to the cluster log accessed
the cluster transporter without using any locks. Because of
this, it was theoretically possible that these signals could be
interefered with by heartbeat signals if both were sent at the
same time, causing the ndb_restore messages
to be corrupted.
(Bug#45646)
Problems could arise when using
VARCHAR columns
whose size was greater than 341 characters and which used the
utf8_unicode_ci collation. In some cases,
this combination of conditions could cause certain queries and
OPTIMIZE TABLE statements to
crash mysqld.
(Bug#45053)
An internal NDB API buffer was not properly initialized.
(Bug#44977)
When a data node had written its GCI marker to the first page of
a megabyte, and that node was later killed during restart after
having processed that page (marker) but before completing a LCP,
the data node could fail with filesystem errors.
(Bug#44952) See also Bug#42564, Bug#44291.
The warning message Possible bug in
Dbdih::execBLOCK_COMMIT_ORD ... could sometimes
appear in the cluster log. This warning is obsolete, and has
been removed.
(Bug#44563)
In some cases, OPTIMIZE TABLE on
an NDB table did not free any
DataMemory .
(Bug#43683)
If the cluster crashed during the execution of a
CREATE LOGFILE GROUP statement,
the cluster could not be restarted afterwards.
(Bug#36702) See also Bug#34102. Disk Data: Partitioning:
An NDBCLUSTER table created with a
very large value for the MAX_ROWS option
could — if this table was dropped and a new table with
fewer partitions, but having the same table ID, was created
— cause ndbd to crash when performing a
system restart. This was because the server attempted to examine
each partition whether or not it actually existed.
(Bug#45154) Disk Data:
If the value set in the config.ini file for
FileSystemPathDD ,
FileSystemPathDataFiles , or
FileSystemPathUndoFiles was identical to the
value set for FileSystemPath , that parameter
was ignored when starting the data node with
--initial option. As a result, the Disk Data
files in the corresponding directory were not removed when
performing an initial start of the affected data node or data
nodes.
(Bug#46243) Disk Data:
During a checkpoint, restore points are created for both the
on-disk and in-memory parts of a Disk Data table. Under certain
rare conditions, the in-memory restore point could include or
exclude a row that should have been in the snapshot. This would
later lead to a crash during or following recovery.
(Bug#41915) See also Bug#47832.
17.7.3.11. Changes in MySQL Cluster NDB 6.3.25 (5.1.34-ndb-6.3.25) (25 May 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.34 (see Section C.1.15, “Changes in MySQL 5.1.34 (02 April 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Two new server status variables
Ndb_scan_count and
Ndb_pruned_scan_count have
been introduced.
Ndb_scan_count gives the
number of scans executed since the cluster was last started.
Ndb_pruned_scan_count gives
the number of scans for which
NDBCLUSTER was able to use
partition pruning. Together, these variables can be used to help
determine in the MySQL server whether table scans are pruned by
NDBCLUSTER .
(Bug#44153)
The ndb_config utility program can now
provide an offline dump of all MySQL Cluster configuration
parameters including information such as default and permitted
values, brief description, and applicable section of the
config.ini file. A dump in text format is
produced when running ndb_config with the new
--configinfo option, and in XML format when the
options --configinfo --xml are used together.
For more information and examples, see
Section 17.4.6, “ndb_config — Extract MySQL Cluster Configuration Information”.
Bugs fixed: Important Change: Partitioning:
User-defined partitioning of an
NDBCLUSTER table without any
primary key sometimes failed, and could cause
mysqld to crash.
Now, if you wish to create an
NDBCLUSTER table with user-defined
partitioning, the table must have an explicit primary key, and
all columns listed in the partitioning expression must be part
of the primary key. The hidden primary key used by the
NDBCLUSTER storage engine is not
sufficient for this purpose. However, if the list of columns is
empty (that is, the table is defined using PARTITION BY
[LINEAR] KEY() ), then no explicit primary key is
required.
This change does not effect the partitioning of tables using any
storage engine other than
NDBCLUSTER .
(Bug#40709) Important Change:
Previously, the configuration parameter
NoOfReplicas had no default value. Now the
default for NoOfReplicas is 2, which is the
recommended value in most settings.
(Bug#44746) Packaging:
The pkg installer for MySQL Cluster on
Solaris did not perform a complete installation due to an
invalid directory reference in the post-install script.
(Bug#41998)
When ndb_config could not find the file
referenced by the --config-file option, it
tried to read my.cnf instead, then failed
with a misleading error message.
(Bug#44846)
When a data node was down so long that its most recent local
checkpoint depended on a global checkpoint that was no longer
restorable, it was possible for it to be unable to use optimized
node recovery when being restarted later.
(Bug#44844) See also Bug#26913.
ndb_config
--xml
did not output any entries for the HostName
parameter. In addition, the default listed for
MaxNoOfFiles was outside the allowed range of
values.
(Bug#44749) See also Bug#44685, Bug#44746.
The output of ndb_config
--xml
did not provide information about all sections of the
configuration file.
(Bug#44685) See also Bug#44746, Bug#44749.
Inspection of the code revealed that several assignment
operators (= ) were used in place of
comparison operators (== ) in
DbdihMain.cpp .
(Bug#44567) See also Bug#44570.
It was possible for NDB API applications to insert corrupt data
into the database, which could subquently lead to data node
crashes. Now, stricter checking is enforced on input data for
inserts and updates.
(Bug#44132)
ndb_restore failed when trying to restore
data on a big-endian machine from a backup file created on a
little-endian machine.
(Bug#44069)
The file ndberror.c contained a C++-style
comment, which caused builds to fail with some C compilers.
(Bug#44036)
When trying to use a data node with an older version of the
management server, the data node crashed on startup.
(Bug#43699)
In some cases, data node restarts during a system restart could
fail due to insufficient redo log space.
(Bug#43156)
NDBCLUSTER did not build correctly
on Solaris 9 platforms.
(Bug#39080) See also Bug#39036, Bug#39038.
ndb_restore --print_data did
not handle DECIMAL columns
correctly.
(Bug#37171)
The output of ndbd --help
did not provide clear information about the program's
--initial and --initial-start
options.
(Bug#28905)
It was theoretically possible for the value of a nonexistent
column to be read as NULL , rather than
causing an error.
(Bug#27843) Disk Data:
This fix supercedes and improves on an earlier fix made for this
bug in MySQL 5.1.18.
(Bug#24521) Cluster Replication:
A failure when setting up replication events could lead to
subsequent data node failures.
(Bug#44915)
17.7.3.12. Changes in MySQL Cluster NDB 6.3.24 (5.1.32-ndb-6.3.24) (09 April 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.24 from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.32-ndb-6.3.24/.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.32 (see Section C.1.17, “Changes in MySQL 5.1.32 (14 February 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: Cluster Replication:
If data node failed during an event creation operation, there
was a slight risk that a surviving data node could send an
invalid table reference back to NDB, causing the operation to
fail with a false Error 723 (No such
table). This could take place when a data node
failed as a mysqld process was setting up
MySQL Cluster Replication.
(Bug#43754) Cluster API:
Partition pruning did not work correctly for queries involving
multiple range scans.
As part of the fix for this issue, several improvements have
been made in the NDB API, including the addition of a new
NdbScanOperation::getPruned() method, a new
variant of NdbIndexScanOperation::setBound() ,
and a new Ndb::PartitionSpec data structure.
For more information about these changes, see
NdbScanOperation::getPruned() ,
NdbIndexScanOperation::setBound , and
The PartitionSpec Structure.
(Bug#37934)
TransactionDeadlockDetectionTimeout values
less than 100 were treated as 100. This could cause scans to
time out unexpectedly.
(Bug#44099)
A race condition could occur when a data node failed to restart
just before being included in the next global checkpoint. This
could cause other data nodes to fail.
(Bug#43888)
TimeBetweenLocalCheckpoints was measured from
the end of one local checkpoint to the beginning of the next,
rather than from the beginning of one LCP to the beginning of
the next. This meant that the time spent performing the LCP was
not taken into account when determining the
TimeBetweenLocalCheckpoints interval, so that
LCPs were not started often enough, possibly causing data nodes
to run out of redo log space prematurely.
(Bug#43567)
Using indexes containing variable-sized columns could lead to
internal errors when the indexes were being built.
(Bug#43226)
When a data node process had been killed after allocating a node
ID, but before making contact with any other data node
processes, it was not possible to restart it due to a node ID
allocation failure.
This issue could effect either ndbd or
ndbmtd processes.
(Bug#43224) This regression was introduced by Bug#42973.
Some queries using combinations of logical and comparison
operators on an indexed column in the WHERE
clause could fail with the error Got error 4541
'IndexBound has no bound information' from
NDBCLUSTER.
(Bug#42857)
ndb_restore crashed when trying to restore a
backup made to a MySQL Cluster running on a platform having
different endianness from that on which the original backup was
taken.
(Bug#39540)
When aborting an operation involving both an insert and a
delete, the insert and delete were aborted separately. This was
because the transaction coordinator did not know that the
operations affected on same row, and, in the case of a
committed-read (tuple or index) scan, the abort of the insert
was performed first, then the row was examined after the insert
was aborted but before the delete was aborted. In some cases,
this would leave the row in a inconsistent state. This could
occur when a local checkpoint was performed during a backup.
This issue did not affect primary ley operations or scans that
used locks (these are serialized).
After this fix, for ordered indexes, all operations that follow
the operation to be aborted are now also aborted.
Disk Data:
When a log file group had an undo log file whose size was too
small, restarting data nodes failed with Read
underflow errors.
As a result of this fix, the minimum allowed
INTIAL_SIZE for an undo log file is now
1M (1 megabyte).
(Bug#29574) Cluster Replication:
When creating or altering a table an
NdbEventOperation is created by the
mysqld process to monitor the table for
subsequent logging in the binlog. If this happened during a node
restart there was a chance that the reference count on this
event operation object could be incorrect, which could lead to
an assert in debug MySQL Cluster builds.
(Bug#43752) Cluster API:
If the largest offset of a
RecordSpecification
used for an
NdbRecord
object was for the NULL bits (and thus not a
column), this offset was not taken into account when calculating
the size used for the RecordSpecification .
This meant that the space for the NULL bits
could be overwritten by key or other information.
(Bug#43891) Cluster API:
BIT columns created using the
native NDB API format that were not created as nullable could
still sometimes be overwritten, or cause other columns to be
overwritten.
This issue did not effect tables having
BIT columns created using the
mysqld format (always used by MySQL Cluster SQL nodes).
(Bug#43802) Cluster API:
The default NdbRecord structures created by
NdbDictionary could have overlapping null
bits and data fields.
(Bug#43590) Cluster API:
When performing insert or write operations,
NdbRecord allows key columns to be specified
in both the key record and in the attribute record. Only one key
column value for each key column should be sent to the NDB
kernel, but this was not guaranteed. This is now ensured as
follows: For insert and write operations, key column values are
taken from the key record; for scan takeover update operations,
key column values are taken from the attribute record.
(Bug#42238) Cluster API:
Ordered index scans using NdbRecord formerly
expressed a BoundEQ range as separate lower
and upper bounds, resulting in 2 copies of the column values
being sent to the NDB kernel.
Now, when a range is specified by
NdbScanOperation::setBound() , the passed
pointers, key lengths, and inclusive bits are compared, and only
one copy of the equal key columns is sent to the kernel. This
makes such operations more efficient, as half the amount of
KeyInfo is now sent for a
BoundEQ range as before.
(Bug#38793)
17.7.3.13. Changes in MySQL Cluster NDB 6.3.23 (5.1.32-ndb-6.3.23) (24 February 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.23 from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.32-ndb-6.3.23/.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.32 (see Section C.1.17, “Changes in MySQL 5.1.32 (14 February 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: Performance:
Updates of the SYSTAB_0 system table to
obtain a unique identifier did not use transaction hints for
tables having no primary key. In such cases the NDB kernel used
a cache size of 1. This meant that each insert into a table not
having a primary key required an update of the corresponding
SYSTAB_0 entry, creating a potential
performance bottleneck.
With this fix, inserts on NDB tables without
primary keys can be under some conditions be performed up to
100% faster than previously.
(Bug#39268) Packaging:
Packages for MySQL Cluster were missing the
libndbclient.so and
libndbclient.a files.
(Bug#42278) Partitioning:
Executing ALTER TABLE ... REORGANIZE
PARTITION on an
NDBCLUSTER table having only one
partition caused mysqld to crash.
(Bug#41945) See also Bug#40389.
Backup IDs greater than 231 were not
handled correctly, causing negative values to be used in backup
directory names and printouts.
(Bug#43042)
When using ndbmtd, NDB kernel threads could
hang while trying to start the data nodes with
LockPagesInMainMemory set to 1.
(Bug#43021)
When using multiple management servers and starting several API
nodes (possibly including one or more SQL nodes) whose
connectstrings listed the management servers in different order,
it was possible for 2 API nodes to be assigned the same node ID.
When this happened it was possible for an API node not to get
fully connected, consequently producing a number of errors whose
cause was not easily recognizable.
(Bug#42973)
ndb_error_reporter worked correctly only with
GNU tar. (With other versions of
tar, it produced empty archives.)
(Bug#42753)
Triggers on NDBCLUSTER tables
caused such tables to become locked.
(Bug#42751) See also Bug#16229, Bug#18135.
Given a MySQL Cluster containing no data (that is, whose data
nodes had all been started using --initial , and
into which no data had yet been imported) and having an empty
backup directory, executing START BACKUP with
a user-specified backup ID caused the data nodes to crash.
(Bug#41031)
In some cases, NDB did not check
correctly whether tables had changed before trying to use the
query cache. This could result in a crash of the debug MySQL
server.
(Bug#40464) Disk Data:
It was not possible to add an in-memory column online to a table
that used a table-level or column-level STORAGE
DISK option. The same issue prevented ALTER
ONLINE TABLE ... REORGANIZE PARTITION from working on
Disk Data tables.
(Bug#42549) Disk Data:
Creating a Disk Data tablespace with a very large extent size
caused the data nodes to fail. The issue was observed when using
extent sizes of 100 MB and larger.
(Bug#39096) Disk Data:
Trying to execute a CREATE LOGFILE
GROUP statement using a value greater than
150M for UNDO_BUFFER_SIZE
caused data nodes to crash.
As a result of this fix, the upper limit for
UNDO_BUFFER_SIZE is now
600M ; attempting to set a higher value now
fails gracefully with an error.
(Bug#34102) See also Bug#36702. Disk Data:
When attempting to create a tablespace that already existed, the
error message returned was Table or index with given
name already exists.
(Bug#32662) Disk Data:
Using a path or filename longer than 128 characters for Disk
Data undo log files and tablespace data files caused a number of
issues, including failures of CREATE
LOGFILE GROUP , ALTER LOGFILE
GROUP , CREATE
TABLESPACE , and ALTER
TABLESPACE statements, as well as crashes of
management nodes and data nodes.
With this fix, the maximum length for path and file names used
for Disk Data undo log files and tablespace data files is now
the same as the maximum for the operating system.
(Bug#31769, Bug#31770, Bug#31772) Disk Data:
Attempting to perform a system restart of the cluster where
there existed a logfile group without and undo log files caused
the data nodes to crash.
Note
While issuing a CREATE LOGFILE
GROUP statement without an ADD
UNDOFILE option fails with an error in the MySQL
server, this situation could arise if an SQL node failed
during the execution of a valid CREATE
LOGFILE GROUP statement; it is also possible to
create a logfile group without any undo log files using the
NDB API.
(Bug#17614) Cluster Replication:
Being disconnected from the cluster while setting up the binary
log caused mysqld to hang or crash.
(Bug#43045) Cluster Replication:
Primary key updates on MyISAM and
InnoDB tables failed to replicate
to NDBCLUSTER tables.
(Bug#42921) Cluster API:
Some error messages from ndb_mgmd contained
newline (\n ) characters. This could break the
MGM API protocol, which uses the newline as a line separator.
(Bug#43104) Cluster API:
When using an ordered index scan without putting all key columns
in the read mask, this invalid use of the NDB API went
undetected, which resulted in the use of uninitialized memory.
(Bug#42591)
17.7.3.14. Changes in MySQL Cluster NDB 6.3.22 (5.1.31-ndb-6.3.22) (09 February 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This is a source-only release. You can obtain the GPL source
code for MySQL Cluster NDB 6.3.22 from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.31-ndb-6.3.22/.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.31 (see Section C.1.19, “Changes in MySQL 5.1.31 (19 January 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
New options are introduced for ndb_restore
for determining which tables or databases should be restored:
--include-tables and
--include-databases can be used to restore
specific tables or databases.
--exclude-tables and
--exclude-databases can be used to exclude
the specified tables or databases from being restored.
For more information about these options, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#40429)
Bugs fixed:
When performing more than 32 index or tuple scans on a single
fragment, the scans could be left hanging. This caused
unnecessary timeouts, and in addition could possibly lead to a
hang of an LCP.
(Bug#42559)
A data node failure that occurred between calls to
NdbIndexScanOperation::readTuples(SF_OrderBy)
and NdbTransaction::Execute() was not
correctly handled; a subsequent call to
nextResult() caused a null pointer to be
deferenced, leading to a segfault in mysqld.
(Bug#42545)
Issuing SHOW GLOBAL STATUS LIKE 'NDB%' before
mysqld had connected to the cluster caused a
segmentation fault.
(Bug#42458)
Data node failures that occurred before all data nodes had
connected to the cluster were not handled correctly, leading to
additional data node failures.
(Bug#42422)
When a cluster backup failed with Error 1304 (Node
node_id1 : Backup request from
node_id2 failed to start), no clear
reason for the failure was provided.
As part of this fix, MySQL Cluster now retries backups in the
event of sequence errors.
(Bug#42354) See also Bug#22698.
Issuing SHOW ENGINE
NDBCLUSTER STATUS on an SQL node before the management
server had connected to the cluster caused
mysqld to crash.
(Bug#42264)
17.7.3.15. Changes in MySQL Cluster NDB 6.3.21 (5.1.31-ndb-6.3.21) (19 January 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
Note
MySQL Cluster NDB 6.3.21 was withdrawn due to issues
discovered after its release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.31 (see Section C.1.19, “Changes in MySQL 5.1.31 (19 January 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Change:
Formerly, when the management server failed to create a
transporter for a data node connection,
net_write_timeout seconds
elapsed before the data node was actually allowed to disconnect.
Now in such cases the disconnection occurs immediately.
(Bug#41965) See also Bug#41713.
It is now possible while in Single User Mode to restart all data
nodes using ALL RESTART in the management
client. Restarting of individual nodes while in Single User Mode
remains disallowed.
(Bug#31056)
Formerly, when using MySQL Cluster Replication, records for
“empty” epochs — that is, epochs in which no
changes to NDBCLUSTER data or
tables took place — were inserted into the
ndb_apply_status and
ndb_binlog_index tables on the slave even
when --log-slave-updates was
disabled. Beginning with MySQL Cluster NDB 6.2.16 and MySQL
Cluster NDB 6.3.13 this was changed so that these
“empty” eopchs were no longer logged. However, it
is now possible to re-enable the older behavior (and cause
“empty” epochs to be logged) by using the
--ndb-log-empty-epochs option. For more
information, see Section 16.1.3.3, “Replication Slave Options and Variables”.
See also Bug#37472.
Bugs fixed:
A maximum of 11 TUP scans were allowed in
parallel.
(Bug#42084)
Trying to execute an
ALTER ONLINE TABLE
... ADD COLUMN statement while inserting rows into the
table caused mysqld to crash.
(Bug#41905)
If the master node failed during a global checkpoint, it was
possible in some circumstances for the new master to use an
incorrect value for the global checkpoint index. This could
occur only when the cluster used more than one node group.
(Bug#41469)
API nodes disconnected too agressively from cluster when data
nodes were being restarted. This could sometimes lead to the API
node being unable to access the cluster at all during a rolling
restart.
(Bug#41462)
It was not possible to perform online upgrades from a MySQL
Cluster NDB 6.2 release to MySQL Cluster NDB 6.3.8 or a later
MySQL Cluster NDB 6.3 release.
(Bug#41435)
Cluster log files were opened twice by internal log-handling
code, resulting in a resource leak.
(Bug#41362)
An abort path in the DBLQH kernel block
failed to release a commit acknowledgement marker. This meant
that, during node failure handling, the local query handler
could be added multiple times to the marker record which could
lead to additional node failures due an array overflow.
(Bug#41296)
During node failure handling (of a data node other than the
master), there was a chance that the master was waiting for a
GCP_NODEFINISHED signal from the failed node
after having received it from all other data nodes. If this
occurred while the failed node had a transaction that was still
being committed in the current epoch, the master node could
crash in the DBTC kernel block when
discovering that a transaction actually belonged to an epoch
which was already completed.
(Bug#41295)
Issuing EXIT in the management client
sometimes caused the client to hang.
(Bug#40922)
In the event that a MySQL Cluster backup failed due to file
permissions issues, conflicting reports were issued in the
management client.
(Bug#34526)
If all data nodes were shut down, MySQL clients were unable to
access NDBCLUSTER tables and data
even after the data nodes were restarted, unless the MySQL
clients themselves were restarted.
(Bug#33626) Disk Data:
Starting a cluster under load such that Disk Data tables used
most of the undo buffer could cause data node failures.
The fix for this bug also corrected an issue in the
LGMAN kernel block where the amount of free
space left in the undo buffer was miscalculated, causing buffer
overruns. This could cause records in the buffer to be
overwritten, leading to problems when restarting data nodes.
(Bug#28077) Cluster Replication:
Sometimes, when using the --ndb_log_orig
option, the orig_epoch and
orig_server_id columns of the
ndb_binlog_index table on the slave contained
the ID and epoch of the local server instead.
(Bug#41601) Cluster API:
mgmapi.h contained constructs which only
worked in C++, but not in C.
(Bug#27004)
17.7.3.16. Changes in MySQL Cluster NDB 6.3.20 (5.1.30-ndb-6.3.20) (17 December 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
Obtaining MySQL Cluster NDB 6.3.20.
You can download MySQL Cluster NDB 6.3.20 source code and
binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.30 (see Section C.1.20, “Changes in MySQL 5.1.30 (14 November 2008 General Availability)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
If a transaction was aborted during the handling of a data node
failure, this could lead to the later handling of an API node
failure not being completed.
(Bug#41214)
Issuing SHOW TABLES repeatedly could cause
NDBCLUSTER tables to be dropped.
(Bug#40854)
Statements of the form UPDATE ... ORDER BY ...
LIMIT run against
NDBCLUSTER tables failed to update
all matching rows, or failed with the error Can't
find record in
'table_name '.
(Bug#40081)
Start phase reporting was inconsistent between the management
client and the cluster log.
(Bug#39667)
Status messages shown in the management client when restarting a
management node were inappropriate and misleading. Now, when
restarting a management node, the messages displayed are as
follows, where node_id is the
management node's node ID:
ndb_mgm> node_id RESTART
Shutting down MGM node node_id for restart
Node node_id is being restarted
ndb_mgm>
(Bug#29275) Partitioning:
A query on a user-partitioned table caused MySQL to crash, where
the query had the following characteristics:
The query's WHERE clause referenced
an indexed column that was also in the partitioning key.
The query's WHERE clause included a
value found in the partition.
The query's WHERE clause used the
< or <>
operators to compare with the indexed column's value
with a constant.
The query used an ORDER BY clause, and
the same indexed column was used in the ORDER
BY clause.
The ORDER BY clause used an explcit or
implicit ASC sort priority.
Two examples of such a query are given here, where
a represents an indexed column used in the
table's partitioning key:
SELECT * FROM table WHERE a < constant ORDER BY a;
SELECT * FROM table WHERE a <> constant ORDER BY a;
This bug was introduced in MySQL Cluster NDB 6.3.19.
(Bug#40954) This regression was introduced by Bug#30573, Bug#33257, Bug#33555. Replication:
Issuing the statement CHANGE MASTER TO ...
MASTER_HEARTBEAT_PERIOD =
period using a value for
period outside the permitted range
caused the slave to crash.
(Bug#39077) Disk Data:
This improves on a previous fix for this issue that was made in
MySQL Cluster 6.3.8.
(Bug#37116) See also Bug#29186. Cluster API:
When creating a scan using an NdbScanFilter
object, it was possible to specify conditions against a
BIT column, but the correct rows were not
returned when the scan was executed.
As part of this fix, 4 new comparison operators have been
implemented for use with scans on BIT
columns:
COL_AND_MASK_EQ_MASK
COL_AND_MASK_NE_MASK
COL_AND_MASK_EQ_ZERO
COL_AND_MASK_NE_ZERO
For more information about these operators, see
The NdbScanFilter::BinaryCondition Type.
Equivalent methods are now also defined for
NdbInterpretedCode ; for more information, see
NdbInterpretedCode Bitwise Comparison Operations.
(Bug#40535)
17.7.3.17. Changes in MySQL Cluster NDB 6.3.19 (5.1.29-ndb-6.3.19) (21 November 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.29 (see Section C.1.21, “Changes in MySQL 5.1.29 (11 October 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster API: Important Change:
MGM API applications exited without raising any errors if the
connection to the management server was lost. The fix for this
issue includes two changes:
The MGM API now provides its own
SIGPIPE handler to catch the
“broken pipe” error that occurs when writing
to a closed or reset socket. This means that MGM API now
behaves the same as NDB API in this regard.
A new function
ndb_mgm_set_ignore_sigpipe() has been
added to the MGM API. This function makes it possible to
bypass the SIGPIPE handler provded by
the MGM API.
(Bug#40498) Cluster Replication: Important Note:
This release of MySQL Cluster derives in part from MySQL 5.1.29,
where the default value for the
--binlog-format option changed to
STATEMENT . That change does
not affect this or future MySQL Cluster NDB
6.x releases, where the default value for this option remains
MIXED , since MySQL Cluster Replication does
not work with the statement-based format.
(Bug#40586)
When performing an initial start of a data node, fragment log
files were always created sparsely — that is, not all
bytes were written. Now it is possible to override this behavior
using the new InitFragmentLogFiles
configuration parameter.
(Bug#40847)
Bugs fixed: Cluster API:
Failed operations on BLOB and
TEXT columns were not always
reported correctly to the originating SQL node. Such errors were
sometimes reported as being due to timeouts, when the actual
problem was a transporter overload due to insufficient buffer
space.
(Bug#39867, Bug#39879)
Undo logs and data files were created in 32K increments. Now
these files are created in 512K increments, resulting in shorter
creation times.
(Bug#40815)
Redo log creation was very slow on some platforms, causing MySQL
Cluster to start more slowly than necessary with some
combinations of hardware and operating system. This was due to
all write operations being synchronized to disk while creating a
redo log file. Now this synchronization occurs only after the
redo log has been created.
(Bug#40734)
Transaction failures took longer to handle than was necessary.
When a data node acting as transaction coordinator (TC) failed,
the surviving data nodes did not inform the API node initiating
the transaction of this until the failure had been processed by
all protocols. However, the API node needed only to know about
failure handling by the transaction protocol — that is, it
needed to be informed only about the TC takeover process. Now,
API nodes (including MySQL servers acting as cluster SQL nodes)
are informed as soon as the TC takeover is complete, so that it
can carry on operating more quickly.
(Bug#40697)
It was theoretically possible for stale data to be read from
NDBCLUSTER tables when the
transaction isolation level was set to
ReadCommitted .
(Bug#40543)
The LockExecuteThreadToCPU and
LockMaintThreadsToCPU parameters did not work
on Solaris.
(Bug#40521)
SET SESSION ndb_optimized_node_selection = 1
failed with an invalid warning message.
(Bug#40457)
A restarting data node could fail with an error in the
DBDIH kernel block when a local or global
checkpoint was started or triggered just as the node made a
request for data from another data node.
(Bug#40370)
Restoring a MySQL Cluster from a dump made using
mysqldump failed due to a spurious error:
Can't execute the given command because you have
active locked tables or an active transaction.
(Bug#40346)
O_DIRECT was incorrectly disabled when making
MySQL Cluster backups.
(Bug#40205)
Heavy DDL usage caused the mysqld processes
to hang due to a timeout error (NDB
error code 266).
(Bug#39885)
Executing EXPLAIN
SELECT on an NDBCLUSTER
table could cause mysqld to crash.
(Bug#39872)
Events logged after setting ALL CLUSTERLOG
STATISTICS=15 in the management client did not always
include the node ID of the reporting node.
(Bug#39839)
The MySQL Query Cache did not function correctly with
NDBCLUSTER tables containing
TEXT columns.
(Bug#39295)
A segfault in Logger::Log caused
ndbd to hang indefinitely. This fix improves
on an earlier one for this issue, first made in MySQL Cluster
NDB 6.2.16 and MySQL Cluster NDB 6.3.17.
(Bug#39180) See also Bug#38609.
Memory leaks could occur in handling of strings used for storing
cluster metadata and providing output to users.
(Bug#38662)
A duplicate key or other error raised when inserting into an
NDBCLUSTER table caused the current
transaction to abort, after which any SQL statement other than a
ROLLBACK
failed. With this fix, the
NDBCLUSTER storage engine now
performs an implicit rollback when a transaction is aborted in
this way; it is no longer necessary to issue an explicit
ROLLBACK
statement, and the next statement that is issued automatically
begins a new transaction.
Note
It remains necessary in such cases to retry the complete
transaction, regardless of which statement caused it to be
aborted.
(Bug#32656) See also Bug#47654.
Error messages for NDBCLUSTER error
codes 1224 and 1227 were missing.
(Bug#28496) Partitioning:
Dropping or creating an index on a partitioned table managed by
the InnoDB Plugin locked the table.
(Bug#37453) Disk Data:
Issuing concurrent CREATE TABLESPACE ,
ALTER TABLESPACE , CREATE LOGFILE
GROUP , or ALTER LOGFILE GROUP
statements on separate SQL nodes caused a resource leak that led
to data node crashes when these statements were used again
later.
(Bug#40921) Disk Data:
Disk-based variable-length columns were not always handled like
their memory-based equivalents, which could potentially lead to
a crash of cluster data nodes.
(Bug#39645) Disk Data:
O_SYNC was incorrectly disabled on platforms
that do not support O_DIRECT . This issue was
noted on Solaris but could have affected other platforms not
having O_DIRECT capability.
(Bug#34638) Cluster API:
The MGM API reset error codes on management server handles
before checking them. This meant that calling an MGM API
function with a null handle caused applications to crash.
(Bug#40455) Cluster API:
It was not always possible to access parent objects directly
from NdbBlob ,
NdbOperation , and
NdbScanOperation objects. To alleviate this
problem, a new getNdbOperation() method has
been added to NdbBlob and new
getNdbTransaction() methods have been added to
NdbOperation and
NdbScanOperation . In addition, a const
variant of NdbOperation::getErrorLine() is
now also available.
(Bug#40242) Cluster API:
NdbScanOperation::getBlobHandle() failed when
used with incorrect column names or numbers.
(Bug#40241) Cluster API:
The MGM API function ndb_mgm_listen_event()
ignored bind addresses.
As part of this fix, it is now possible to specify bind
addresses in connectstrings. See
Section 17.3.2.3, “The MySQL Cluster Connectstring”, for more
information.
(Bug#38473) Cluster API:
The NDB API example programs included in MySQL Cluster source
distributions failed to compile.
(Bug#37491) See also Bug#40238.
17.7.3.18. Changes in MySQL Cluster NDB 6.3.18 (5.1.28-ndb-6.3.18) (03 October 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.3 release.
Obtaining MySQL Cluster NDB 6.3.
You can download the latest MySQL Cluster NDB 6.3 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.28 (see Section C.1.22, “Changes in MySQL 5.1.28 (28 August 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
When a transaction included a multi-row insert to an
NDBCLUSTER table that caused a
constraint violation, the transaction failed to roll back.
(Bug#395638)
Starting the MySQL Server with the
--ndbcluster option plus an
invalid command-line option (for example, using
mysqld --ndbcluster
--foobar ) caused it to hang while shutting down the
binlog thread.
(Bug#39635)
Dropping and then re-creating a database on one SQL node caused
other SQL nodes to hang.
(Bug#39613)
Setting a low value of MaxNoOfLocalScans
(< 100) and performing a large number of (certain) scans
could cause the Transaction Coordinator to run out of scan
fragment records, and then crash. Now when this resource is
exhausted, the cluster returns Error 291 (Out of
scanfrag records in TC (increase MaxNoOfLocalScans))
instead.
(Bug#39549)
Creating a unique index on an
NDBCLUSTER table caused a memory
leak in the NDB subscription
manager (SUMA ) which could lead to mysqld
hanging, due to the fact that the resource shortage was not
reported back to the NDB kernel
correctly.
(Bug#39518) See also Bug#39450.
Embedded libmysqld with
NDB did not drop table events.
(Bug#39450)
Unique identifiers in tables having no primary key were not
cached. This fix has been observed to increase the efficiency of
INSERT operations on such tables
by as much as 50%.
(Bug#39267)
When restarting a data node, an excessively long shutodwn
message could cause the node process to crash.
(Bug#38580)
After a forced shutdown and initial restart of the cluster, it
was possible for SQL nodes to retain .frm
files corresponding to NDBCLUSTER
tables that had been dropped, and thus to be unaware that these
tables no longer existed. In such cases, attempting to re-create
the tables using CREATE TABLE IF NOT EXISTS
could fail with a spurious Table ... doesn't
exist error.
(Bug#37921)
A statement of the form DELETE FROM
table WHERE
primary_key =value
or UPDATE table WHERE
primary_key =value
where there was no row whose primary key column had the stated
value appeared to succeed, with the
server reporting that 1 row had been changed.
This issue was only known to affect MySQL Cluster NDB 6.3.11 and
later NDB 6.3 versions.
(Bug#37153) Cluster Replication:
In some cases, dropping a database on the master could cause
table logging to fail on the slave, or, when using a debug
build, could cause the slave mysqld to fail
completely.
(Bug#39404) Cluster API:
Passing a value greater than 65535 to
NdbInterpretedCode::add_val() and
NdbInterpretedCode::sub_val() caused these
methods to have no effect.
(Bug#39536)
17.7.3.19. Changes in MySQL Cluster NDB 6.3.17 (5.1.27-ndb-6.3.17) (28 August 2008)
This is a new release, fixing recently discovered bugs in
previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
Previously, MySQL Cluster NDB 6.3 releases were source-only
releases which required compiling and installing using the
instructions found in Section 2.3, “MySQL Installation Using a Source Distribution”, and
in Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. Beginning
with MySQL Cluster NDB 6.3.17, binaries built from NDB 6.3
sources are also available. You can download the latest MySQL
Cluster NDB 6.2 source code and binaries for supported
platforms from http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.3 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.27 (see Section C.1.23, “Changes in MySQL 5.1.27 (Not released)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: Packaging:
Support for the InnoDB storage engine was
missing from the GPL source releases. An updated GPL source
tarball
mysql-5.1.27-ndb-6.3.17-innodb.tar.gz which
includes code for building InnoDB can be
found on
the
MySQL FTP site.
MgmtSrvr::allocNodeId() left a mutex locked
following an Ambiguity for node if %d
error.
(Bug#39158)
An invalid path specification caused
mysql-test-run.pl to fail.
(Bug#39026)
During transactional coordinator takeover (directly after node
failure), the LQH finding an operation in the
LOG_COMMIT state sent an
LQH_TRANS_CONF signal twice, causing the TC
to fail.
(Bug#38930)
An invalid memory access caused the management server to crash
on Solaris Sparc platforms.
(Bug#38628)
A segfault in Logger::Log caused
ndbd to hang indefinitely.
(Bug#38609)
ndb_mgmd failed to start on older Linux
distributions (2.4 kernels) that did not support e-polling.
(Bug#38592)
ndb_mgmd sometimes performed unnecessary
network I/O with the client. This in combination with other
factors led to long-running threads that were attempting to
write to clients that no longer existed.
(Bug#38563)
ndb_restore failed with a floating point
exception due to a division by zero error when trying to restore
certain data files.
(Bug#38520)
A failed connection to the management server could cause a
resource leak in ndb_mgmd.
(Bug#38424)
Failure to parse configuration parameters could cause a memory
leak in the NDB log parser.
(Bug#38380)
Renaming an NDBCLUSTER table on one
SQL node, caused a trigger on this table to be deleted on
another SQL node.
(Bug#36658)
Attempting to add a UNIQUE INDEX twice to an
NDBCLUSTER table, then deleting
rows from the table could cause the MySQL Server to crash.
(Bug#35599)
ndb_restore failed when a single table was
specified.
(Bug#33801)
GCP_COMMIT did not wait for transaction
takeover during node failure. This could cause
GCP_SAVE_REQ to be executed too early. This
could also cause (very rarely) replication to skip rows.
(Bug#30780) Cluster Replication:
During a parallel node restart, the starting nodes could
(sometimes) incorrectly synchronize subscriptions among
themselves. Instead, this synchronization now takes place only
among nodes that have actually (completely) started.
(Bug#38767) Cluster API:
Support for Multi-Range Read index scans using the old API
(using, for example,
NdbIndexScanOperation::setBound() or
NdbIndexScanOperation::end_of_bound() ) were
dropped in MySQL Cluster NDB 6.2. This functionality is restored
in MySQL Cluster NDB 6.3 beginning with 6.3.17, but remains
unavailable in MySQL Cluster NDB 6.2. Both MySQL Cluster NDB 6.2
and 6.3 support Multi-Range Read scans via the
NdbRecord API.
(Bug#38791) Cluster API:
The NdbScanOperation::readTuples() method
could be called multiple times without error.
(Bug#38717) Cluster API:
Certain Multi-Range Read scans involving IS
NULL and IS NOT NULL comparisons
failed with an error in the NDB
local query handler.
(Bug#38204) Cluster API:
Problems with the public headers prevented
NDB applications from being built
with warnings turned on.
(Bug#38177) Cluster API:
Creating an NdbScanFilter object using an
NdbScanOperation object that had not yet had
its readTuples() method called resulted in a
crash when later attempting to use the
NdbScanFilter .
(Bug#37986) Cluster API:
Executing an NdbRecord interpreted delete
created with an ANYVALUE option caused the
transaction to abort.
(Bug#37672)
17.7.3.20. Changes in MySQL Cluster NDB 6.3.16 (5.1.24-ndb-6.3.16) (27 June 2008)
This is a new source release, fixing recently discovered bugs in
previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This release incorporates all bugfixes and changes made in the
previous MySQL Cluster NDB 6.3 release, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.24 (see Section C.1.26, “Changes in MySQL 5.1.24 (08 April 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Event buffer lag reports are now written to the cluster log.
(Bug#37427)
Added the --no-binlog option for
ndb_restore. When used, this option prevents
information being written to SQL node binary logs from the
restoration of a cluster backup.
(Bug#30452)
Bugs fixed: Cluster API:
Changing the system time on data nodes could cause MGM API
applications to hang and the data nodes to crash.
(Bug#35607)
Failure of a data node could sometimes cause mysqld to crash.
(Bug#37628)
DELETE ... WHERE
unique_index_column =value
deleted the wrong row from the table.
(Bug#37516)
If subscription was terminated while a node was down, the epoch
was not properly acknowledged by that node.
(Bug#37442)
libmysqld failed to wait for the cluster
binlog thread to terminate before exiting.
(Bug#37429)
In rare circumstances, a connection followed by a disconnection
could give rise to a “stale” connection where the
connection still existed but was not seen by the transporter.
(Bug#37338)
Queries against NDBCLUSTER tables
were cached only if autocommit
was in use.
(Bug#36692) Cluster Replication:
Data was written to the binlog with
--log-slave-updates disabled.
(Bug#37472) Cluster API:
When some operations succeeded and some failed following a call
to NdbTransaction::execute(Commit,
AO_IgnoreOnError) , a race condition could cause
spurious occurrences of NDB API Error 4011 (Internal
error).
(Bug#37158) Cluster API:
Creating a table on an SQL node, then starting an NDB API
application that listened for events from this table, then
dropping the table from an SQL node, prevented data node
restarts.
(Bug#32949, Bug#37279) Cluster API:
A buffer overrun in NdbBlob::setValue()
caused erroneous results on Mac OS X.
(Bug#31284)
17.7.3.21. Changes in MySQL Cluster NDB 6.3.15 (5.1.24-ndb-6.3.15) (30 May 2008)
This is a new source release, fixing recently discovered bugs in
previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This release incorporates all bugfixes and changes made in the
previous MySQL Cluster NDB 6.3 release, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.24 (see Section C.1.26, “Changes in MySQL 5.1.24 (08 April 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
In certain rare situations, ndb_size.pl
could fail with the error Can't use string
("value ") as a HASH ref while "strict
refs" in use.
(Bug#43022)
Under some circumstances, a failed CREATE
TABLE could mean that subsequent
CREATE TABLE statements caused
node failures.
(Bug#37092)
A fail attempt to create an NDB
table could in some cases lead to resource leaks or cluster
failures.
(Bug#37072)
Attempting to create a native backup of
NDB tables having a large number of
NULL columns and data could lead to node
failures.
(Bug#37039)
Checking of API node connections was not efficiently handled.
(Bug#36843)
Attempting to delete a nonexistent row from a table containing a
TEXT or
BLOB column within a transaction
caused the transaction to fail.
(Bug#36756) See also Bug#36851.
If the combined total of tables and indexes in the cluster was
greater than 4096, issuing START BACKUP
caused data nodes to fail.
(Bug#36044)
Where column values to be compared in a query were of the
VARCHAR or
VARBINARY types,
NDBCLUSTER passed a value padded to
the full size of the column, which caused unnecessary data to be
sent to the data nodes. This also had the effect of wasting CPU
and network bandwidth, and causing condition pushdown to be
disabled where it could (and should) otherwise have been
applied.
(Bug#35393)
When dropping a table failed for any reason (such as when in
single user mode) then the corresponding
.ndb file was still removed.
Replication:
When flushing tables, there was a slight chance that the flush
occurred between the processing of one table map event and the
next. Since the tables were opened one by one, subsequent
locking of tables would cause the slave to crash. This problem
was observed when replicating
NDBCLUSTER or
InnoDB tables, when executing multi-table
updates, and when a trigger or a stored routine performed an
(additional) insert on a table so that two tables were
effectively being inserted into in the same statement.
(Bug#36197) Cluster API:
Ordered index scans were not pruned correctly where a
partitioning key was specified with an EQ-bound.
(Bug#36950) Cluster API:
When an insert operation involving
BLOB data was attempted on a row
which already existed, no duplicate key error was correctly
reported and the transaction is incorrectly aborted. In some
cases, the existing row could also become corrupted.
(Bug#36851) See also Bug#26756. Cluster API:
NdbApi.hpp depended on
ndb_global.h , which was not actually
installed, causing the compilation of programs that used
NdbApi.hpp to fail.
(Bug#35853)
17.7.3.22. Changes in MySQL Cluster NDB 6.3.14 (5.1.24-ndb-6.3.14) (11 May 2008)
This is a new source release, fixing recently discovered bugs in
previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This release incorporates all bugfixes and changes made in the
previous MySQL Cluster NDB 6.3 release, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.24 (see Section C.1.26, “Changes in MySQL 5.1.24 (08 April 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
SET GLOBAL ndb_extra_logging caused
mysqld to crash.
(Bug#36547)
A race condition caused by a failure in epoll handling could
cause data nodes to fail.
(Bug#36537)
Under certain rare circumstances, the failure of the new master
node while attempting a node takeover would cause takeover
errors to repeat without being resolved.
(Bug#36199, Bug#36246, Bug#36247, Bug#36276)
When more than one SQL node connected to the cluster at the same
time, creation of the mysql.ndb_schema table
failed on one of them with an explicit Table
exists error, which was not necessary.
(Bug#35943)
mysqld failed to start after running
mysql_upgrade.
(Bug#35708)
Notification of a cascading master node failures could sometimes
not be transmitted correctly (that is, transmission of the
NF_COMPLETEREP signal could fail), leading to
transactions hanging and timing out
(NDB error 4012), scans hanging,
and failure of the management server process.
(Bug#32645)
If an API node disconnected and then reconnected during Start
Phase 8, then the connection could be “blocked”
— that is, the QMGR kernel block failed
to detect that the API node was in fact connected to the
cluster, causing issues with the
NDB Subscription Manager
(SUMA ).
NDB error 1427 (Api node
died, when SUB_START_REQ reached node) was
incorrectly classified as a schema error rather than a temporary
error.
Cluster Replication:
Performing SELECT ... FROM
mysql.ndb_apply_status before the
mysqld process had connected to the cluster
failed, and caused this table never to be created.
(Bug#36123) Cluster API:
Accesing the debug version of libndbclient
via dlopen() resulted in a segmentation
fault.
(Bug#35927) Cluster API:
Attempting to pass a nonexistent column name to the
equal() and setValue()
methods of NdbOperation caused NDB API
applications to crash. Now the column name is checked, and an
error is returned in the event that the column is not found.
(Bug#33747) Cluster API:
Relocation errors were encountered when trying to compile NDB
API applications on a number of platforms, including 64-bit
Linux. As a result, libmysys ,
libmystrings , and libdbug
have been changed from normal libraries to “noinst”
libtool helper libraries. They are no longer
installed as separate libraries; instead, all necessary symbols
from these are added directly to
libndbclient . This means that NDB API
programs now need to be linked only using
-lndbclient .
(Bug#29791)
17.7.3.23. Changes in MySQL Cluster NDB 6.3.13 (5.1.24-ndb-6.3.13) (10 April 2008)
This is a new source release, fixing recently discovered bugs in
previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This release incorporates all bugfixes and changes made in the
previous MySQL Cluster NDB 6.3 release, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.24 (see Section C.1.26, “Changes in MySQL 5.1.24 (08 April 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: Important Change:
mysqld_safe now traps Signal 13
(SIGPIPE ) so that this signal no longer kills
the MySQL server process.
(Bug#33984)
Node or system restarts could fail due an unitialized variable
in the DTUP kernel block. This issue was
found in MySQL Cluster NDB 6.3.11.
(Bug#35797)
If an error occured while executing a statement involving a
BLOB or
TEXT column of an
NDB table, a memory leak could
result.
(Bug#35593)
It was not possible to determine the value used for the
--ndb-cluster-connection-pool option in the
mysql client. Now this value is reported as a
system status variable.
(Bug#35573)
The ndb_waiter utility wrongly calculated
timeouts.
(Bug#35435)
A SELECT on a table with a
nonindexed, large VARCHAR column
which resulted in condition pushdown on this column could cause
mysqld to crash.
(Bug#35413)
ndb_restore incorrectly handled some
datatypes when applying log files from backups.
(Bug#35343)
In some circumstances, a stopped data node was handled
incorrectly, leading to redo log space being exhausted following
an initial restart of the node, or an initial or partial restart
of the cluster (the wrong CGI might be used in such cases). This
could happen, for example, when a node was stopped following the
creation of a new table, but before a new LCP could be executed.
(Bug#35241)
SELECT ... LIKE ... queries yielded incorrect
results when used on NDB tables. As
part of this fix, condition pushdown of such queries has been
disabled; re-enabling it is expected to be done as part of a
later, permanent fix for this issue.
(Bug#35185)
ndb_mgmd reported errors to
STDOUT rather than to
STDERR .
(Bug#35169)
Nested Multi-Range Read scans failed when the second Multi-Range
Read released the first read's unprocessed operations,
sometimes leading to an SQL node crash.
(Bug#35137)
In some situations, a problem with synchronizing checkpoints
between nodes could cause a system restart or a node restart to
fail with Error 630 during restore of TX.
(Bug#34756)
A node failure during an initial node restart followed by
another node start could cause the master data node to fail,
because it incorrectly gave the node permission to start even if
the invalidated node's LCP was still running.
(Bug#34702)
When a secondary index on a
DECIMAL column was used to
retrieve data from an NDB table, no
results were returned even if the target table had a matched
value in the column that was defined with the secondary index.
(Bug#34515)
An UPDATE on an
NDB table that set a new value for
a unique key column could cause subsequent queries to fail.
(Bug#34208)
If a data node in one node group was placed in the “not
started” state (using
node_id RESTART
-n ), it was not possible to stop a data node in a
different node group.
(Bug#34201)
Numerous NDBCLUSTER test failures
occurred in builds compiled using icc on IA64
platforms.
(Bug#31239)
If a START BACKUP command was issued while
ndb_restore was running, the backup being
restored could be overwritten.
(Bug#26498)
REPLACE statements did not work
correctly with NDBCLUSTER tables
when all columns were not explicitly listed.
(Bug#22045)
CREATE TABLE and
ALTER TABLE statements using
ENGINE=NDB or
ENGINE=NDBCLUSTER caused
mysqld to fail on Solaris 10 for x86
platforms.
(Bug#19911) Replication:
A CHANGE MASTER TO statement with
no MASTER_HEARTBEAT_PERIOD option failed to
reset the heartbeat period to its default value.
(Bug#34686) Cluster Replication:
In some cases, when updating only one or some columns in a
table, the complete row was written to the binary log instead of
only the updated column or columns, even when
ndb_log_updated_only was set to 1.
(Bug#35208) Cluster Replication:
Enabling the ndb_wait_connected
system variable caused the server to wait for a partial
connection plus an additional 3 seconds for a complete
connection to the cluster. This could lead to issues with
setting up the binary log.
(Bug#34757) Cluster API:
Closing a scan before it was executed caused the application to
segfault.
(Bug#36375) Cluster API:
Using NDB API applications from older MySQL Cluster versions
with libndbclient from newer ones caused the
cluster to fail.
(Bug#36124) Cluster API:
Some ordered index scans could return tuples out of order.
(Bug#35908) Cluster API:
Scans having no bounds set were handled incorrectly.
(Bug#35876) Cluster API:
NdbScanFilter::getNdbOperation() , which was
inadvertently removed in MySQL Cluster NDB 6.3.11, has been
restored.
(Bug#35854)
Enabling the read_only system
variable while autocommit mode
was enabled caused SELECT
statements for transactional storage engines to fail.
(Bug#35732)
Executing a FLUSH
PRIVILEGES statement after creating a temporary table
in the mysql database with the same name as
one of the MySQL system tables caused the server to crash.
Note
While it is possible to shadow a system table in this way,
the temporary table exists only for the current user and
connection, and does not effect any user privileges.
(Bug#33275)
17.7.3.24. Changes in MySQL Cluster NDB 6.3.12 (5.1.23-ndb-6.3.12) (05 April 2008)
MySQL Cluster NDB 6.3.12 was pulled due to issues discovered
shortly after its release, and is no longer available. Users of
MySQL Cluster NDB 6.3.10 and earlier MySQL Cluster NDB 6.3
releases should upgrade to MySQL Cluster NDB 6.3.13 or later.
For information about bugfixes and feature enhancements that
were originally scheduled to appear for the first time in this
release, see
Section 17.7.3.23, “Changes in MySQL Cluster NDB 6.3.13 (5.1.24-ndb-6.3.13) (10 April 2008)”.
17.7.3.25. Changes in MySQL Cluster NDB 6.3.11 (5.1.23-ndb-6.3.11) (28 March 2008)
This release was pulled due to issues discovered shortly after
its release, and is no longer available. Users of MySQL Cluster
NDB 6.3.10 and earlier MySQL Cluster NDB 6.3 releases should
upgrade to MySQL Cluster NDB 6.3.13 or later.
For information about bugfixes and feature enhancements that
were originally scheduled to appear for the first time in this
release, see
Section 17.7.3.23, “Changes in MySQL Cluster NDB 6.3.13 (5.1.24-ndb-6.3.13) (10 April 2008)”.
17.7.3.26. Changes in MySQL Cluster NDB 6.3.10 (5.1.23-ndb-6.3.10) (17 February 2008)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
Due to the reduction of the number of local checkpoints from 3
to 2 in MySQL Cluster NDB 6.3.8, a data node using ndbd from
MySQL Cluster NDB 6.3.8 or later started using a file system
from an earlier version could incorrectly invalidate local
checkpoints too early during the startup process, causing the
node to fail.
(Bug#34596)
17.7.3.27. Changes in MySQL Cluster NDB 6.3.9 (5.1.23-ndb-6.3.9) (12 February 2008)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
Cluster failures could sometimes occur when performing more than
three parallel takeovers during node restarts or system
restarts. This affected MySQL Cluster NDB
6.3.x releases only.
(Bug#34445)
Upgrades of a cluster using while a
DataMemory setting in excess of 16 GB caused
data nodes to fail.
(Bug#34378)
Performing many SQL statements on
NDB tables while in
autocommit mode caused a memory
leak in mysqld.
(Bug#34275)
In certain rare circumstances, a race condition could occur
between an aborted insert and a delete leading a data node
crash.
(Bug#34260)
Multi-table updates using ordered indexes during handling of
node failures could cause other data nodes to fail.
(Bug#34216)
When configured with NDB support,
MySQL failed to compile using gcc 4.3 on
64bit FreeBSD systems.
(Bug#34169)
The failure of a DDL statement could sometimes lead to node
failures when attempting to execute subsequent DDL statements.
(Bug#34160)
Extremely long SELECT statements
(where the text of the statement was in excess of 50000
characters) against NDB tables
returned empty results.
(Bug#34107)
When configured with NDB support,
MySQL failed to compile on 64bit FreeBSD systems.
(Bug#34046)
Statements executing multiple inserts performed poorly on
NDB tables having
AUTO_INCREMENT columns.
(Bug#33534)
The ndb_waiter utility polled
ndb_mgmd excessively when obtaining the
status of cluster data nodes.
(Bug#32025) See also Bug#32023.
Transaction atomicity was sometimes not preserved between reads
and inserts under high loads.
(Bug#31477)
Having tables with a great many columns could cause Cluster
backups to fail.
(Bug#30172) Cluster Replication: Disk Data:
Statements violating unique keys on Disk Data tables (such as
attempting to insert NULL into a NOT
NULL column) could cause data nodes to fail. When the
statement was executed from the binlog, this could also result
in failure of the slave cluster.
(Bug#34118) Disk Data:
Updating in-memory columns of one or more rows of Disk Data
table, followed by deletion of these rows and re-insertion of
them, caused data node failures.
(Bug#33619) Cluster Replication:
Setting
--replicate-ignore-db=mysql
caused the mysql.ndb_apply_status table not
to be replicated, breaking Cluster Replication.
(Bug#28170)
17.7.3.28. Changes in MySQL Cluster NDB 6.3.8 (5.1.23-ndb-6.3.8) (29 January 2008 General Availability)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster API: Important Change:
Because NDB_LE_MemoryUsage.page_size_kb shows
memory page sizes in bytes rather than kilobytes, it has been
renamed to page_size_bytes . The name
page_size_kb is now deprecated and thus
subject to removal in a future release, although it currently
remains supported for reasons of backward compatibility. See
The Ndb_logevent_type Type, for more information
about NDB_LE_MemoryUsage .
(Bug#30271)
ndb_restore now supports basic
attribute promotion; that is, data from a
column of a given type can be restored to a column using a
“larger” type. For example, Cluster backup data
taken from a SMALLINT column can
be restored to a MEDIUMINT ,
INT , or
BIGINT column.
For more information, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
Now only 2 local checkpoints are stored, rather than 3 as in
previous MySQL Cluster versions. This lowers disk space
requirements and reduces the size and number of redo log files
needed.
The mysqld option
--ndb-batch-size has been added. This allows
for controlling the size of batches used for running
transactions.
Node recovery can now be done in parallel, rather than
sequentially, which can result in much faster recovery times.
Persistence of NDB tables can now
be controlled using the session variables
ndb_table_temporary and
ndb_table_no_logging .
ndb_table_no_logging causes
NDB tables not to be checkpointed
to disk; ndb_table_temporary does the same,
and in addition, no schema files are created.
OPTIMIZE TABLE can now be
interrupted. This can be done, for example, by killing the SQL
thread performing the OPTIMIZE operation.
Bugs fixed: Disk Data: Important Change:
It is no longer possible on 32-bit systems to issue statements
appearing to create Disk Data log files or data files greater
than 4 GB in size. (Trying to create log files or data files
larger than 4 GB on 32-bit systems led to unrecoverable data
node failures; such statements now fail with
NDB error 1515.)
(Bug#29186) Replication:
The code implementing heartbeats did not check for possible
errors in some circumstances; this kept the dump thread hanging
while waiting for heartbeats loop even though the slave was no
longer connected.
(Bug#33332)
High numbers of insert operations, delete operations, or both
could cause NDB error 899
(Rowid already allocated) to occur
unnecessarily.
(Bug#34033)
A periodic failure to flush the send buffer by the
NDB TCP transporter could cause a
unnecessary delay of 10 ms between operations.
(Bug#34005)
DROP TABLE did not free all data
memory. This bug was observed in MySQL Cluster NDB 6.3.7 only.
(Bug#33802)
A race condition could occur (very rarely) when the release of a
GCI was followed by a data node failure.
(Bug#33793)
Some tuple scans caused the wrong memory page to be accessed,
leading to invalid results. This issue could affect both
in-memory and Disk Data tables.
(Bug#33739)
A failure to initialize an internal variable led to sporadic
crashes during cluster testing.
(Bug#33715)
The server failed to reject properly the creation of an
NDB table having an unindexed
AUTO_INCREMENT column.
(Bug#30417)
Issuing an
INSERT ...
ON DUPLICATE KEY UPDATE concurrently with or following
a TRUNCATE TABLE statement on an
NDB table failed with
NDB error 4350
Transaction already aborted.
(Bug#29851)
The Cluster backup process could not detect when there was no
more disk space and instead continued to run until killed
manually. Now the backup fails with an appropriate error when
disk space is exhausted.
(Bug#28647)
It was possible in config.ini to define
cluster nodes having node IDs greater than the maximum allowed
value.
(Bug#28298)
Under some circumstances, a recovering data node did not use its
own data, instead copying data from another node even when this
was not required. This in effect bypassed the optimized node
recovery protocol and caused recovery times to be unnecessarily
long.
(Bug#26913) Cluster Replication:
Consecutive DDL statements involving tables
(CREATE TABLE ,
ALTER TABLE , and
DROP TABLE ) could be executed so
quickly that previous DDL statements upon which they depended
were not yet written in the binary log.
For example, if DROP TABLE foo was issued
immediately following CREATE TABLE foo , the
DROP statement could fail because the
CREATE had not yet been recorded.
(Bug#34006) Cluster Replication:
ndb_restore -e restored excessively large
values to the ndb_apply_status table's
epoch column when restoring to a MySQL
Cluster version supporting Micro-GCPs from an older version that
did not support these.
A workaround when restoring to MySQL Cluster releases supporting
micro-GCPs previous to MySQL Cluster NDB 6.3.8 is to perform a
32-bit shift on the epoch column values to
reduce them to their proper size.
(Bug#33406) Cluster API:
Transactions containing inserts or reads would hang during
NdbTransaction::execute() calls made from NDB
API applications built against a MySQL Cluster version that did
not support micro-GCPs accessing a later version that supported
micro-GCPs. This issue was observed while upgrading from MySQL
Cluster NDB 6.1.23 to MySQL Cluster NDB 6.2.10 when the API
application built against the earlier version attempted to
access a data node already running the later version, even after
disabling micro-GCPs by setting
TimeBetweenEpochs equal to 0.
(Bug#33895) Cluster API:
When reading a BIT(64) value using
NdbOperation:getValue() , 12 bytes were
written to the buffer rather than the expected 8 bytes.
(Bug#33750)
17.7.3.29. Changes in MySQL Cluster NDB 6.3.7 (5.1.23-ndb-6.3.7) (19 December 2007)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Compressed local checkpoints and backups are now supported,
resulting in a space savings of 50% or more over uncompressed
LCPs and backups. Compression of these can be enabled in the
config.ini file using the two new data node
configuration parameters CompressedLCP and
CompressedBackup , respectively.
OPTIMIZE TABLE is now supported
for NDBCLUSTER tables, subject to
the following limitations:
Only in-memory tables are supported.
OPTIMIZE still has no effect on Disk Data
tables.
Only variable-length columns are supported. However, you can
force columns defined using fixed-length data types to be
dynamic using the ROW_FORMAT or
COLUMN_FORMAT option with a
CREATE TABLE or
ALTER TABLE statement.
Memory reclaimed from an NDB table
using OPTIMIZE is generally available to the
cluster, and not confined to the table from which it was
recovered, unlike the case with memory freed using
DELETE .
The performance of OPTIMIZE on
NDB tables can be regulated by
adjusting the value of the
ndb_optimization_delay system variable.
It is now possible to cause statements occurring within the same
transaction to be run as a batch by setting the session variable
transaction_allow_batching to
1 or ON .
Note
To use this feature,
autocommit must be disabled.
Bugs fixed: Partitioning:
When partition pruning on an NDB
table resulted in an ordered index scan spanning only one
partition, any descending flag for the scan was wrongly
discarded, causing ORDER BY DESC to be
treated as ORDER BY ASC ,
MAX() to be handled incorrectly, and similar
problems.
(Bug#33061)
When all data and SQL nodes in the cluster were shut down
abnormally (that is, other than by using STOP
in the cluster management client), ndb_mgm
used excessive amounts of CPU.
(Bug#33237)
When using micro-GCPs, if a node failed while preparing for a
global checkpoint, the master node would use the wrong GCI.
(Bug#32922)
Under some conditions, performing an ALTER
TABLE on an NDBCLUSTER
table failed with a Table is full error,
even when only 25% of DataMemory was in use
and the result should have been a table using less memory (for
example, changing a VARCHAR(100) column to
VARCHAR(80) ).
(Bug#32670) Cluster Replication: Replication:
Where a table being replicated had a
TEXT or
BLOB column, an
UPDATE on the master that did not
refer explicitly to this column in the WHERE
clause stopped the SQL thread on the slave with Error
in Write_rows event: row application failed. Got error 4288
'Blob handle for column not available' from
NDBCLUSTER.
(Bug#30674) Cluster Replication:
Creating the mysql.ndb_replication table with
the wrong number of columns for the primary key caused
mysqld to crash. Now a CREATE TABLE
[mysql.]ndb_replication statement that is invalid for
this reason fails with the error Bad schema for
mysql.ndb_replication table. Message: Wrong number of primary
keys, expected number .
(Bug#33159)
17.7.3.30. Changes in MySQL Cluster NDB 6.3.6 (5.1.22-ndb-6.3.6) (08 November 2007)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Note:
MySQL Cluster NDB 6.2 and 6.3 source archives are now available
in separate commercial and GPL versions. Due to licensing
concerns, previous MySQL Cluster NDB 6.2 and 6.3 source archives
were removed from the FTP site.
Unnecessary reads when performing a primary key or unique key
update have been reduced, and in some cases, eliminated. (It is
almost never necessary to read a record prior to an update, the
lone exception to this being when a primary key is updated,
since this requires a delete followed by an insert, which must
be prepared by reading the record.) Depending on the number of
primary key and unique key lookups that are performed per
transaction, this can yield a considerable improvement in
performance.
Batched operations are now better supported for
DELETE and
UPDATE . (UPDATE
WHERE... and muliple
DELETE .)
Introduced the
Ndb_execute_count status
variable, which measures the number of round trips made by
queries to the NDB kernel.
Bugs fixed:
An insert or update with combined range and equality constraints
failed when run against an NDB
table with the error Got unknown error from
NDB. An example of such a statement would be
UPDATE t1 SET b = 5 WHERE a IN (7,8) OR a >=
10; .
(Bug#31874)
An error with an if statement in
sql/ha_ndbcluster.cc could potentially lead
to an infinite loop in case of failure when working with
AUTO_INCREMENT columns in
NDB tables.
(Bug#31810)
The NDB storage engine code was not
safe for strict-alias optimization in gcc
4.2.1.
(Bug#31761)
ndb_restore displayed incorrect backup file
version information. This meant (for example) that, when
attempting to restore a backup made from a MySQL 5.1.22 cluster
to a MySQL Cluster NDB 6.3.3 cluster, the restore process failed
with the error Restore program older than backup
version. Not supported. Use new restore program.
(Bug#31723)
Following an upgrade, ndb_mgmd would fail
with an ArbitrationError.
(Bug#31690)
The NDB management client command
node_id REPORT
MEMORY provided no output when
node_id was the node ID of a
management or API node. Now, when this occurs, the management
client responds with Node
node_id : is not a data
node .
(Bug#29485)
Performing DELETE operations
after a data node had been shut down could lead to inconsistent
data following a restart of the node.
(Bug#26450)
UPDATE IGNORE could sometimes fail on
NDB tables due to the use of
unitialized data when checking for duplicate keys to be ignored.
(Bug#25817) Cluster Replication: Replication:
Slave batching did not work correctly with
UPDATE statements.
(Bug#31787) Cluster Replication: Replication:
A node failure during replication could lead to buckets out of
order; now active subscribers are checked for, rather than empty
buckets.
(Bug#31701) Cluster Replication:
Updates performed unnecessary writes to the primary keys of the
rows being updated.
(Bug#31841) Cluster Replication:
When the master mysqld crashed or was
restarted, no LOST_EVENTS entry was made in
the binlog.
(Bug#31484) See also Bug#21494.
17.7.3.32. Changes in MySQL Cluster NDB 6.3.4 (5.1.22-ndb-6.3.4) (15 October 2007)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Incompatible Change:
The --ndb_optimized_node_selection startup
option for mysqld now allows a wider range of
values and corresponding behaviors for SQL nodes when selecting
a transaction coordinator.
You should be aware that the default value and behavior as well
as the value type used for this option have changed, and that
you may need to update the setting used for this option in your
my.cnf file prior to upgrading
mysqld. See
Section 5.1.4, “Server System Variables”, for more information.
Cluster Replication: Replication:
A replication heartbeat mechanism has been added to facilitate
monitoring. This provides an alternative to checking log files,
making it possible to detect in real time when a slave has
failed.
Configuration of heartbeats is done via a new
MASTER_HEARTBEAT_PERIOD =
interval clause for the
CHANGE MASTER TO statement (see
Section 12.6.2.1, “CHANGE MASTER TO Syntax”); monitoring can be done by
checking the values of the status variables
Slave_heartbeat_period and
Slave_received_heartbeats (see
Section 5.1.7, “Server Status Variables”).
The addition of replication heartbeats addresses a number of
issues:
(Bug#20435, Bug#29309, Bug#30932) Replication:
On MySQL replication slaves having multiple network interfaces,
it is now possible to set which interface to use for connecting
to the master. This can be done by using either the
mysqld startup option
--master-bind or the
MASTER_BIND='interface '
clause in a CHANGE MASTER TO
statement.
(Bug#25939) Cluster Replication:
A new configuration parameter
TimeBetweenEpochsTimeout allows a timeout to
be set for time between epochs. For more information, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
(Bug#31276) Cluster Replication:
Support for a new conflict resolution function
NDB$OLD() has been added for handling
simultaneous updates in multi-master and circular replication
setups. A new status variable
Ndb_conflict_fn_old tracks the
number of times that updates are prevented from being applied
due to this type of conflict resolution. See
Section 17.6.11, “MySQL Cluster Replication Conflict Resolution”,
for more information.
Additional checks were implemented to catch unsupported online
ALTER TABLE operations. Currently
it is not possible to reorder columns or to change the storage
engine used for a table via online ALTER
TABLE .
Some redundant checks made during online creation of indexes
were removed.
A --bind-address option has been
added to a number of MySQL client programs:
mysql, mysqldump,
mysqladmin, mysqlbinlog,
mysqlcheck, mysqlimport,
and mysqlshow. This is for use on a computer
having multiple network interfaces, and allows you to choose
which interface is used to connect to the MySQL server.
Bugs fixed:
It was possible in some cases for a node group to be
“lost” due to missed local checkpoints following a
system restart.
(Bug#31525)
NDB tables having names containing
nonalphanumeric characters (such as
“$ ”) were not discovered
correctly.
(Bug#31470)
A node failure during a local checkpoint could lead to a
subsequent failure of the cluster during a system restart.
(Bug#31257)
A cluster restart could sometimes fail due to an issue with
table IDs.
(Bug#30975)
Transaction timeouts were not handled well in some
circumstances, leading to excessive number of transactions being
aborted unnecessarily.
(Bug#30379)
In some cases, the cluster managment server logged entries
multiple times following a restart of mgmd.
(Bug#29565)
ndb_mgm --help did not
display any information about the -a option.
(Bug#29509)
An interpreted program of sufficient size and complexity could
cause all cluster data nodes to shut down due to buffer
overruns.
(Bug#29390)
The cluster log was formatted inconsistently and contained
extraneous newline characters.
(Bug#25064)
A transaction was not aborted following the failure of
statement.
(Bug#31320)
Online ALTER operations involving a column
whose data type has an implicit default value left behind
temporary .frm files, causing subsequent
DROP DATABASE statements to fail.
(Bug#31097)
Errors could sometimes occur during an online ADD
COLUMN under load.
(Bug#31082)
Transactions were committed prematurely when
LOCK
TABLE and SET autocommit = 0 were
used together.
(Bug#30996)
The mysqld_safe script contained a syntax
error.
(Bug#30624)
17.7.3.33. Changes in MySQL Cluster NDB 6.3.3 (5.1.22-ndb-6.3.3) (20 September 2007)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Mapping of NDB error codes to MySQL
storage engine error codes has been improved.
(Bug#28423) Cluster Replication: Replication:
A server status variable ndb_conflict_fn_max
now provides a count of the number of times that conflict
resolution for MySQL Cluster Replication has been applied.
See
Section 17.6.11, “MySQL Cluster Replication Conflict Resolution”,
for more information.
Bugs fixed: Partitioning:
EXPLAIN
PARTITIONS reported partition usage by queries on
NDB tables according to the
standard MySQL hash function than the hash function used in the
NDB storage engine.
(Bug#29550)
Attempting to restore a backup made on a cluster host using one
endian to a machine using the other endian could cause the
cluster to fail.
(Bug#29674)
The description of the --print option provided
in the output from ndb_restore --help
was incorrect.
(Bug#27683)
Restoring a backup made on a cluster host using one endian to a
machine using the other endian failed for
BLOB and
DATETIME columns.
(Bug#27543, Bug#30024)
Errors could sometimes occur during an online ADD
COLUMN under load.
(Bug#31082)
17.7.3.34. Changes in MySQL Cluster NDB 6.3.2 (5.1.22-ndb-6.3.2) (07 September 2007)
This is a new Beta development release, fixing recently
discovered bugs in previous MySQL Cluster NDB 6.3 releases.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.3 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Online ADD COLUMN , ADD
INDEX , and DROP INDEX
operations can now be performed explicitly for
NDB tables, as well as online
renaming of tables and columns for
NDB and MyISAM
tables — that is, without copying or locking of the
affected tables — using ALTER ONLINE
TABLE .
Indexes can also be created and dropped online using
CREATE INDEX and
DROP INDEX , respectively, using
the ONLINE keyword.
You can force operations that would otherwise be performed
online to be done offline using the OFFLINE
keyword.
See Section 12.1.7, “ALTER TABLE Syntax”,
Section 12.1.13, “CREATE INDEX Syntax”, and
Section 12.1.24, “DROP INDEX Syntax”, for more information.
It is now possible to control whether fixed-width or
variable-width storage is used for a given column of an
NDB table by means of the
COLUMN_FORMAT specifier as part of the
column's definition in a CREATE
TABLE or ALTER TABLE
statement.
It is also possible to control whether a given column of an
NDB table is stored in memory or on
disk, using the STORAGE specifier as part of
the column's definition in a CREATE
TABLE or ALTER TABLE
statement.
For permitted values and other information about
COLUMN_FORMAT and STORAGE ,
see Section 12.1.17, “CREATE TABLE Syntax”.
A new cluster management server startup option
--bind-address makes it possible
to restrict management client connections to
ndb_mgmd to a single host and port. For more
information, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
Cluster Replication:
Multi-way replication failover and recovery for
NDB is facilitated with the
introduction of the --ndb-log-orig option. When
mysqld is started with this option, the
originating server ID and epoch of each binlog event is recorded
in the mysql.ndb_binlog_index table, which
now contains two additional columns
orig_server_id and
orig_epoch for storing this information. In
such cases, a single epoch on a slave may be represented by
multiple rows in the slave's
ndb_binlog_index table, one for each epoch as
it originated from a master.
Cluster Replication:
The protocol for handling global checkpoints has been changed.
It is now possible to control how often the GCI number is
updated, and how often global checkpoints are written to disk,
using the TimeBetweenEpochs configuration
parameter. This improves the reliability and performance of
MySQL Cluster Replication.
GCPs handled using the new protocol are sometimes referred to as
“micro-GCPs”.
For more information, see
TimeBetweenEpochs
.
Bugs fixed:
When an NDB event was left behind
but the corresponding table was later recreated and received a
new table ID, the event could not be dropped.
(Bug#30877)
When creating an NDB table with a column that
has COLUMN_FORMAT = DYNAMIC , but the table
tiself uses ROW_FORMAT=FIXED , the table is
considered dynamic, but any columns for which the row format is
unspecified default to FIXED . Now in such
cases the server issues the warning Row format FIXED
incompatible with dynamic attribute
column_name .
(Bug#30276)
An insufficiently descriptive and potentially misleading Error
4006 (Connect failure - out of connection
objects...) was produced when either of the
following two conditions occurred:
There were no more transaction records in the transaction
coordinator
An NDB object in the NDB API
was initialized with insufficient parallelism
Separate error messages are now generated for each of these two
cases.
(Bug#11313)
For micro-GCPs, the assignment of “fake” CGI events
no longer cause buckets to be sent out of order. Now, when
assigning a GCI to a non-GCI event (that is, creating a
pseudo-GCI or “fake” CGI), the GCI that is to
arrive is always initiated, even if no known GCI exists, which
could occur in the event of a node failure.
(Bug#30884)
17.7.3.36. Changes in MySQL Cluster NDB 6.3.0 (5.1.19-ndb-6.3.0) (02 July 2007 Beta)
This is the first development release for MySQL Cluster NDB 6.3,
based on version 6.3 of the
NDBCLUSTER storage engine.
Obtaining MySQL Cluster NDB 6.3.
This is a source-only release, which you must compile and
install using the instructions found in
Section 2.3, “MySQL Installation Using a Source Distribution”, and in
Section 17.2.1, “MySQL Cluster Multi-Computer Installation”. You can
download the GPL source tarball from the MySQL FTP site at
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/.
This Beta release incorporates all bugfixes and feature changes
made in previous MySQL Cluster releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.19 (see Section C.1.31, “Changes in MySQL 5.1.19 (25 May 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Reporting functionality has been significantly enhanced in this
release:
A new configuration parameter
BackupReportFrequency now makes it
possible to cause the management client to provide status
reports at regular intervals as well as for such reports
to be written to the cluster log (depending on cluster
event logging levels). See
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for more
information about this parameter.
A new REPORT command has been added in
the cluster management client. REPORT
BackupStatus allows you to obtain a backup
status report at any time during a backup. REPORT
MemoryUsage reports the current data memory and
index memory used by each data node. For more about the
REPORT command, see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”.
ndb_restore now provides running
reports of its progress when restoring a backup. In
addition, a complete report status report on the backup is
written to the cluster log.
A new configuration parameter ODirect causes
NDB to attempt using
O_DIRECT writes for LCP, backups, and redo
logs, often lowering CPU usage.
Cluster Replication:
This release implements conflict resolution, which makes it
possible to determine on a per-table basis whether or not an
update to a given row on the master should be applied on the
slave. For more information, see
Section 17.6.11, “MySQL Cluster Replication Conflict Resolution”.
17.7.4. Changes in MySQL Cluster NDB 6.2
This section contains change history information for MySQL Cluster
releases based on version 6.2 of the
NDBCLUSTER storage engine.
For an overview of new features added in MySQL Cluster NDB 6.2,
see Section 17.1.4.3, “MySQL CLuster Development in MySQL Cluster NDB 6.2”.
17.7.4.1. Changes in MySQL Cluster NDB 6.2.19 (5.1.41-ndb-6.2.19) (Not yet released)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.41 (see Section C.1.5, “Changes in MySQL 5.1.41 (05 November 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: Important Change:
The --with-ndb-port-base option for
configure did not function correctly, and has
been deprecated. Attempting to use this option produces the
warning Ignoring deprecated option
--with-ndb-port-base.
Beginning with MySQL Cluster NDB 7.1.0, the deprecation warning
itself is removed, and the --with-ndb-port-base
option is simply handled as an unknown and invalid option if you
try to use it.
(Bug#47941) See also Bug#38502. Cluster Replication: Important Change:
In a MySQL Cluster acting as a replication slave and having
multiple SQL nodes, only the SQL node receiving events directly
from the master recorded DDL statements in its binary logs
unless this SQL node also had binary logging enabled; otherwise,
other SQL nodes in the slave cluster failed to log DDL
statements, regardless of their individual
--log-bin settings.
The fix for this issue aligns binary logging of DDL statements
with that of DML statements. In particular, you should take note
of the following:
DDL and DML statements on the master cluster are logged with
the server ID of the server that actually writes the log.
DDL and DML statements on the master cluster are logged by
any attached mysqld that has binary
logging enabled.
Replicated DDL and DML statements on the slave are logged by
any attached mysqld that has both
--log-bin and
--log-slave-updates enabled.
Replicated DDL and DML statements are logged with the server
ID of the original (master) MySQL server by any attached
mysqld that has both
--log-bin and
--log-slave-updates enabled.
Affect on upgrades.
When upgrading from a previous MySQL CLuster release, you
should perform either one of the following:
Upgrade servers that are performing binary logging
before those that are not; do not perform any DDL on
“old” SQL nodes until all SQL nodes have
been upgraded.
Make sure that
--log-slave-updates is
enabled on all SQL nodes performing binary logging prior
to the upgrade, so that all DDL is captured.
Note
Logging of DML statements was not affected by this issue.
(Bug#45756) Packaging:
The pkg installer for MySQL Cluster on
Solaris did not perform a complete installation due to an
invalid directory reference in the post-install script.
(Bug#41998)
When using NoOfReplicas equal to 1 or 2, if
data nodes from one node group were restarted 256 times and
applications were running traffic such that it would encounter
NDB error 1204
(Temporary failure, distribution
changed), the live node in the node group would
crash, causing the cluster to crash as well. The crash occurred
only when the error was encountered on the 256th restart; having
the error on any previous or subsequent restart did not cause
any problems.
(Bug#50930)
If a query on an NDB table compared
a constant string value to a column, and the length of the
string was greater than that of the column, condition pushdown
did not work correctly. (The string was truncated to fit the
column length before being pushed down.) Now in such cases, the
condition is no longer pushed down.
(Bug#49459)
When performing tasks that generated large amounts of I/O (such
as when using ndb_restore), an internal
memory buffer could overflow, causing data nodes to fail with
signal 6.
Subsequent analysis showed that this buffer was not actually
required, so this fix removes it.
(Bug#48861)
Performing intensive inserts and deletes in parallel with a high
scan load could a data node crashes due to a failure in the
DBACC kernel block. This was because checking
for when to perform bucket splits or merges considered the first
4 scans only.
(Bug#48700)
The creation of an ordered index on a table undergoing DDL
operations could cause a data node crash under certain
timing-dependent conditions.
(Bug#48604)
In certain cases, performing very large inserts on
NDB tables when using
ndbmtd caused the memory allocations for
ordered or unique indexes (or both) to be exceeded. This could
cause aborted transactions and possibly lead to data node
failures.
(Bug#48037) See also Bug#48113.
When employing NDB native backup to
back up and restore an empty NDB
table that used a non-sequential
AUTO_INCREMENT value, the
AUTO_INCREMENT value was not restored
correctly.
(Bug#48005)
SHOW CREATE TABLE did not display
the AUTO_INCREMENT value for
NDB tables having
AUTO_INCREMENT columns.
(Bug#47865)
Under some circumstances, when a scan encountered an error early
in processing by the DBTC kernel block (see
The DBTC Block), a node
could crash as a result. Such errors could be caused by
applications sending incorrect data, or, more rarely, by a
DROP TABLE operation executed in
parallel with a scan.
(Bug#47831)
When starting a node and synchronizing tables, memory pages were
allocated even for empty fragments. In certain situations, this
could lead to insufficient memory.
(Bug#47782)
mysqld allocated an excessively large buffer
for handling BLOB values due to
overestimating their size. (For each row, enough space was
allocated to accommodate every
BLOB or
TEXT column value in the result
set.) This could adversely affect performance when using tables
containing BLOB or
TEXT columns; in a few extreme
cases, this issue could also cause the host system to run out of
memory unexpectedly.
(Bug#47574) See also Bug#47572, Bug#47573.
NDBCLUSTER uses a dynamically-allocated
buffer to store BLOB or
TEXT column data that is read
from rows in MySQL Cluster tables.
When an instance of the NDBCLUSTER table
handler was recycled (this can happen due to table definition
cache pressure or to operations such as
FLUSH TABLES or
ALTER TABLE ), if the last row
read contained blobs of zero length, the buffer was not freed,
even though the reference to it was lost. This resulted in a
memory leak.
For example, consider the table defined and populated as shown
here:
CREATE TABLE t (a INT PRIMARY KEY, b LONGTEXT) ENGINE=NDB;
INSERT INTO t VALUES (1, REPEAT('F', 20000));
INSERT INTO t VALUES (2, '');
Now execute repeatedly a SELECT
on this table, such that the zero-length
LONGTEXT row is
last, followed by a FLUSH
TABLES statement (which forces the handler object to
be re-used), as shown here:
SELECT a, length(b) FROM bl ORDER BY a;
FLUSH TABLES;
Prior to the fix, this resulted in a memory leak proportional to
the size of the stored
LONGTEXT value
each time these two statements were executed.
(Bug#47573) See also Bug#47572, Bug#47574.
Large transactions involving joins between tables containing
BLOB columns used excessive
memory.
(Bug#47572) See also Bug#47573, Bug#47574.
A variable was left uninitialized while a data node copied data
from its peers as part of its startup routine; if the starting
node died during this phase, this could lead a crash of the
cluster when the node was later restarted.
(Bug#47505)
NDB stores blob column data in a
separate, hidden table that is not accessible from MySQL. If
this table was missing for some reason (such as accidental
deletion of the file corresponding to the hidden table) when
making a MySQL Cluster native backup, ndb_restore crashed when
attempting to restore the backup. Now in such cases, ndb_restore
fails with the error message Table
table_name has blob column
(column_name ) with missing parts
table in backup instead.
(Bug#47289)
For very large values of MaxNoOfTables +
MaxNoOfAttributes , the calculation for
StringMemory could overflow when creating
large numbers of tables, leading to NDB error 773
(Out of string memory, please modify StringMemory
config parameter), even when
StringMemory was set to
100 (100 percent).
(Bug#47170)
The default value for the StringMemory
configuration parameter, unlike other MySQL Cluster
configuration parameters, was not set in
ndb/src/mgmsrv/ConfigInfo.cpp .
(Bug#47166)
Signals from a failed API node could be received after an
API_FAILREQ signal (see
Operations and Signals)
has been received from that node, which could result in invalid
states for processing subsequent signals. Now, all pending
signals from a failing API node are processed before any
API_FAILREQ signal is received.
(Bug#47039) See also Bug#44607.
Using triggers on NDB tables caused
ndb_autoincrement_prefetch_sz
to be treated as having the NDB kernel's internal default
value (32) and the value for this variable as set on the
cluster's SQL nodes to be ignored.
(Bug#46712)
Full table scans failed to execute when the cluster contained
more than 21 table fragments.
Note
The number of table fragments in the cluster can be calculated
as the number of data nodes, times 8 (that is, times the value
of the internal constant
MAX_FRAG_PER_NODE ), divided by the number
of replicas. Thus, when NoOfReplicas = 1 at
least 3 data nodes were required to trigger this issue, and
when NoOfReplicas = 2 at least 4 data nodes
were required to do so.
(Bug#46490)
Ending a line in the config.ini file with
an extra semicolon character (; ) caused
reading the file to fail with a parsing error.
(Bug#46242)
When combining an index scan and a delete with a primary key
delete, the index scan and delete failed to initialize a flag
properly. This could in rare circumstances cause a data node to
crash.
(Bug#46069)
Problems could arise when using
VARCHAR columns
whose size was greater than 341 characters and which used the
utf8_unicode_ci collation. In some cases,
this combination of conditions could cause certain queries and
OPTIMIZE TABLE statements to
crash mysqld.
(Bug#45053)
Running an ALTER TABLE statement
while an NDB backup was in progress caused
mysqld to crash.
(Bug#44695)
If a node failed while sending a fragmented long signal, the
receiving node did not free long signal assembly resources that
it had allocated for the fragments of the long signal that had
already been received.
(Bug#44607)
When performing auto-discovery of tables on individual SQL
nodes, NDBCLUSTER attempted to overwrite
existing MyISAM .frm
files and corrupted them.
Workaround.
In the mysql client, create a new table
(t2 ) with same definition as the corrupted
table (t1 ). Use your system shell or file
manager to rename the old .MYD file to
the new file name (for example, mv t1.MYD
t2.MYD). In the mysql client,
repair the new table, drop the old one, and rename the new
table using the old file name (for example,
RENAME TABLE t2
TO t1 ).
(Bug#42614)
When starting a cluster with a great many tables, it was
possible for MySQL client connections as well as the slave SQL
thread to issue DML statements against MySQL Cluster tables
before mysqld had finished connecting to the
cluster and making all tables writeable. This resulted in
Table ... is read only errors for clients
and the Slave SQL thread.
This issue is fixed by introducing the
--ndb-wait-setup option for the
MySQL server. This provides a configurable maximum amount of
time that mysqld waits for all
NDB tables to become writeable,
before allowing MySQL clients or the slave SQL thread to
connect.
(Bug#40679) See also Bug#46955.
Running ndb_restore with the
--print or --print_log option
could cause it to crash.
(Bug#40428, Bug#33040)
When building MySQL Cluster, it was possible to configure the
build using --with-ndb-port without supplying a
port number. Now in such cases, configure
fails with an error.
(Bug#38502) See also Bug#47941.
An insert on an NDB table was not
always flushed properly before performing a scan. One way in
which this issue could manifest was that
LAST_INSERT_ID() sometimes failed
to return correct values when using a trigger on an
NDB table.
(Bug#38034)
If the cluster crashed during the execution of a
CREATE LOGFILE GROUP statement,
the cluster could not be restarted afterwards.
(Bug#36702) See also Bug#34102.
Some joins on large NDB tables
having TEXT or
BLOB columns could cause
mysqld processes to leak memory. The joins
did not need to reference the
TEXT or
BLOB columns directly for this
issue to occur.
(Bug#36701)
When the MySQL server SQL mode included
STRICT_TRANS_TABLES , storage
engine warnings and error codes specific to
NDB were returned when errors occurred,
instead of the MySQL server errors and error codes expected by
some programming APIs (such as Connector/J) and applications.
(Bug#35990)
On Mac OS X 10.5, commands entered in the management client
failed and sometimes caused the client to hang, although
management client commands invoked using the
--execute (or
-e ) option from the system shell worked
normally.
For example, the following command failed with an error and hung
until killed manually, as shown here:
ndb_mgm> SHOW
Warning, event thread startup failed, degraded printouts as result, errno=36
^C
However, the same management client command, invoked from the
system shell as shown here, worked correctly:
shell> ndb_mgm -e "SHOW"
(Bug#35751) See also Bug#34438.
When a copying operation exhausted the available space on a data
node while copying large BLOB
columns, this could lead to failure of the data node and a
Table is full error on the SQL node which
was executing the operation. Examples of such operations could
include an ALTER TABLE that
changed an INT column to a
BLOB column, or a bulk insert of
BLOB data that failed due to
running out of space or to a duplicate key error.
(Bug#34583, Bug#48040) See also Bug#41674, Bug#45768.
Trying to insert more rows than would fit into an
NDB table caused data nodes to crash. Now in
such situations, the insert fails gracefully with error 633
Table fragment hash index has reached maximum
possible size.
(Bug#34348)
The error message text for NDB
error code 410 (REDO log files
overloaded...) was truncated.
(Bug#23662) Replication:
When mysqlbinlog
--verbose was used to read a
binary log that had been recorded using the row-based format,
the output for events that updated some but not all columns of
tables was not correct.
(Bug#47323) Replication:
In some cases, a STOP SLAVE
statement could cause the replication slave to crash. This issue
was specific to MySQL on Windows or Macintosh platforms.
(Bug#45238, Bug#45242, Bug#45243, Bug#46013, Bug#46014, Bug#46030) See also Bug#40796. Disk Data:
Inserts of blob column values into a Disk Data table that
exhausted the tablespace resulted in misleading error messages
about rows not being found in the table rather than the expected
error Out of extents, tablespace full.
(Bug#48113) See also Bug#48037, Bug#41674. Disk Data:
A local checkpoint of an empty fragment could cause a crash
during a system restart which was based on that LCP.
(Bug#47832) See also Bug#41915. Disk Data:
Calculation of free space for Disk Data table fragments was
sometimes done incorrectly. This could lead to unnecessary
allocation of new extents even when sufficient space was
available in existing ones for inserted data. In some cases,
this might also lead to crashes when restarting data nodes.
Note
This miscalculation was not reflected in the contents of the
INFORMATION_SCHEMA.FILES table,
as it applied to extents allocated to a fragment, and not to a
file.
(Bug#47072) Disk Data:
If the value set in the config.ini file for
FileSystemPathDD ,
FileSystemPathDataFiles , or
FileSystemPathUndoFiles was identical to the
value set for FileSystemPath , that parameter
was ignored when starting the data node with
--initial option. As a result, the Disk Data
files in the corresponding directory were not removed when
performing an initial start of the affected data node or data
nodes.
(Bug#46243) Disk Data:
Repeatedly creating and then dropping Disk Data tables could
eventually lead to data node failures.
(Bug#45794, Bug#48910) Disk Data:
When a crash occurs due to a problem in Disk Data code, the
currently active page list is printed to
stdout (that is, in one or more
ndb_nodeid _out.log
files). One of these lists could contain an endless loop; this
caused a printout that was effectively never-ending. Now in such
cases, a maximum of 512 entries is printed from each list.
(Bug#42431) Cluster Replication:
When expire_logs_days was set,
the thread performing the purge of the log files could deadlock,
causing all binary log operations to stop.
(Bug#49536) Cluster Replication:
When using multiple active replication channels, it was
sometimes possible that a node group would fail on the slave
cluster, causing the slave cluster to shut down.
(Bug#47935) Cluster Replication:
When recording a binary log using the
--ndb-log-update-as-write and
--ndb-log-updated-only options
(both enabled by default) and later attempting to apply that
binary log with mysqlbinlog, any operations
that were played back from the log but which updated only some
(but not all) columns caused any columns that were not updated
to be reset to their default values.
(Bug#47674) See also Bug#47323, Bug#46662. Cluster Replication:
mysqlbinlog failed to apply correctly a
binary log that had been recorded using
--ndb-log-update-as-write=1 .
(Bug#46662) See also Bug#47323, Bug#47674. Cluster API:
When a DML operation failed due to a uniqueness violation on an
NDB table having more than one
unique index, it was difficult to determine which constraint
caused the failure; it was necessary to obtain an
NdbError object, then decode its
details property, which in could lead to
memory management issues in application code.
To help solve this problem, a new API method
Ndb::getNdbErrorDetail() is added, providing
a well-formatted string containing more precise information
about the index that caused the unque constraint violation. The
following additional changes are also made in the NDB API:
For more information, see
Ndb::getNdbErrorDetail() ,
The NdbError Structure, and
Dictionary::listObjects() .
(Bug#48851) Cluster API:
The NDB API methods Dictionary::listEvents() ,
Dictionary::listIndexes() ,
Dictionary::listObjects() , and
NdbOperation::getErrorLine() formerly had
both const and non-const
variants. The non-const versions of these
methods have been removed. In addition, the
NdbOperation::getBlobHandle() method has been
re-implemented in order to provide consistent internal
semantics.
(Bug#47798) Cluster API:
In some circumstances, if an API node encountered a data node
failure between the creation of a transaction and the start of a
scan using that transaction, then any subsequent calls to
startTransaction() and
closeTransaction() could cause the same
transaction to be started and closed repeatedly.
(Bug#47329) Cluster API:
A duplicate read of a column caused NDB API applications to
crash.
(Bug#45282) Cluster API:
Performing multiple operations using the same primary key within
the same
NdbTransaction::execute()
call could lead to a data node crash.
Note
This fix does not make change the fact that performing
multiple operations using the same primary key within the same
execute() is not supported; because there
is no way to determine the order of such operations, the
result of such combined operations remains undefined.
(Bug#44065) See also Bug#44015. Cluster API:
The error handling shown in the example file
ndbapi_scan.cpp included with the MySQL
Cluster distribution was incorrect.
(Bug#39573) Cluster API:
When using blobs, calling getBlobHandle()
requires the full key to have been set using
equal() , because
getBlobHandle() must access the key for
adding blob table operations. However, if
getBlobHandle() was called without first
setting all parts of the primary key, the application using it
crashed. Now, an appropriate error code is returned instead.
(Bug#28116, Bug#48973) API:
The fix for Bug#24507 could lead in some cases to client
application failures due to a race condition. Now the server
waits for the “dummy” thread to return before
exiting, thus making sure that only one thread can initialize
the POSIX threads library.
(Bug#42850)
On Mac OS X or Windows, sending a SIGHUP
signal to the server or an asynchronous flush (triggered by
flush_time ) caused the server
to crash.
(Bug#47525)
When using the ARCHIVE storage
engine, SHOW TABLE STATUS displayed incorrect
information for Max_data_length ,
Data_length and
Avg_row_length .
(Bug#29203)
Installation of MySQL on Windows would fail to set the correct
location for the character set files, which could lead to
mysqld and mysql failing
to initialize properly.
(Bug#17270)
17.7.4.2. Changes in MySQL Cluster NDB 6.2.18 (5.1.34-ndb-6.2.18) (01 June 2009)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.34 (see Section C.1.15, “Changes in MySQL 5.1.34 (02 April 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: Important Change: Partitioning:
User-defined partitioning of an
NDBCLUSTER table without any
primary key sometimes failed, and could cause
mysqld to crash.
Now, if you wish to create an
NDBCLUSTER table with user-defined
partitioning, the table must have an explicit primary key, and
all columns listed in the partitioning expression must be part
of the primary key. The hidden primary key used by the
NDBCLUSTER storage engine is not
sufficient for this purpose. However, if the list of columns is
empty (that is, the table is defined using PARTITION BY
[LINEAR] KEY() ), then no explicit primary key is
required.
This change does not effect the partitioning of tables using any
storage engine other than
NDBCLUSTER .
(Bug#40709)
An internal NDB API buffer was not properly initialized.
(Bug#44977)
When a data node had written its GCI marker to the first page of
a megabyte, and that node was later killed during restart after
having processed that page (marker) but before completing a LCP,
the data node could fail with filesystem errors.
(Bug#44952) See also Bug#42564, Bug#44291.
Inspection of the code revealed that several assignment
operators (= ) were used in place of
comparison operators (== ) in
DbdihMain.cpp .
(Bug#44567) See also Bug#44570.
It was possible for NDB API applications to insert corrupt data
into the database, which could subquently lead to data node
crashes. Now, stricter checking is enforced on input data for
inserts and updates.
(Bug#44132)
TransactionDeadlockDetectionTimeout values
less than 100 were treated as 100. This could cause scans to
time out unexpectedly.
(Bug#44099)
The file ndberror.c contained a C++-style
comment, which caused builds to fail with some C compilers.
(Bug#44036)
A race condition could occur when a data node failed to restart
just before being included in the next global checkpoint. This
could cause other data nodes to fail.
(Bug#43888)
When trying to use a data node with an older version of the
management server, the data node crashed on startup.
(Bug#43699)
Using indexes containing variable-sized columns could lead to
internal errors when the indexes were being built.
(Bug#43226)
In some cases, data node restarts during a system restart could
fail due to insufficient redo log space.
(Bug#43156)
Some queries using combinations of logical and comparison
operators on an indexed column in the WHERE
clause could fail with the error Got error 4541
'IndexBound has no bound information' from
NDBCLUSTER.
(Bug#42857)
ndb_restore --print_data did
not handle DECIMAL columns
correctly.
(Bug#37171)
The output of ndbd --help
did not provide clear information about the program's
--initial and --initial-start
options.
(Bug#28905)
It was theoretically possible for the value of a nonexistent
column to be read as NULL , rather than
causing an error.
(Bug#27843)
When aborting an operation involving both an insert and a
delete, the insert and delete were aborted separately. This was
because the transaction coordinator did not know that the
operations affected on same row, and, in the case of a
committed-read (tuple or index) scan, the abort of the insert
was performed first, then the row was examined after the insert
was aborted but before the delete was aborted. In some cases,
this would leave the row in a inconsistent state. This could
occur when a local checkpoint was performed during a backup.
This issue did not affect primary ley operations or scans that
used locks (these are serialized).
After this fix, for ordered indexes, all operations that follow
the operation to be aborted are now also aborted.
Disk Data: Partitioning:
An NDBCLUSTER table created with a
very large value for the MAX_ROWS option
could — if this table was dropped and a new table with
fewer partitions, but having the same table ID, was created
— cause ndbd to crash when performing a
system restart. This was because the server attempted to examine
each partition whether or not it actually existed.
(Bug#45154) Disk Data:
During a checkpoint, restore points are created for both the
on-disk and in-memory parts of a Disk Data table. Under certain
rare conditions, the in-memory restore point could include or
exclude a row that should have been in the snapshot. This would
later lead to a crash during or following recovery.
(Bug#41915) See also Bug#47832. Disk Data:
When a log file group had an undo log file whose size was too
small, restarting data nodes failed with Read
underflow errors.
As a result of this fix, the minimum allowed
INTIAL_SIZE for an undo log file is now
1M (1 megabyte).
(Bug#29574) Disk Data:
This fix supercedes and improves on an earlier fix made for this
bug in MySQL 5.1.18.
(Bug#24521) Cluster Replication:
A failure when setting up replication events could lead to
subsequent data node failures.
(Bug#44915) Cluster API:
If the largest offset of a
RecordSpecification
used for an
NdbRecord
object was for the NULL bits (and thus not a
column), this offset was not taken into account when calculating
the size used for the RecordSpecification .
This meant that the space for the NULL bits
could be overwritten by key or other information.
(Bug#43891) Cluster API:
The default NdbRecord structures created by
NdbDictionary could have overlapping null
bits and data fields.
(Bug#43590) Cluster API:
When performing insert or write operations,
NdbRecord allows key columns to be specified
in both the key record and in the attribute record. Only one key
column value for each key column should be sent to the NDB
kernel, but this was not guaranteed. This is now ensured as
follows: For insert and write operations, key column values are
taken from the key record; for scan takeover update operations,
key column values are taken from the attribute record.
(Bug#42238) Cluster API:
Ordered index scans using NdbRecord formerly
expressed a BoundEQ range as separate lower
and upper bounds, resulting in 2 copies of the column values
being sent to the NDB kernel.
Now, when a range is specified by
NdbScanOperation::setBound() , the passed
pointers, key lengths, and inclusive bits are compared, and only
one copy of the equal key columns is sent to the kernel. This
makes such operations more efficient, as half the amount of
KeyInfo is now sent for a
BoundEQ range as before.
(Bug#38793)
17.7.4.3. Changes in MySQL Cluster NDB 6.2.17 (5.1.32-ndb-6.2.17) (19 October 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.32 (see Section C.1.17, “Changes in MySQL 5.1.32 (14 February 2009)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Change:
Formerly, when the management server failed to create a
transporter for a data node connection,
net_write_timeout seconds
elapsed before the data node was actually allowed to disconnect.
Now in such cases the disconnection occurs immediately.
(Bug#41965) See also Bug#41713. Important Change: Replication:
RESET MASTER and
RESET SLAVE now reset the values
shown for Last_IO_Error ,
Last_IO_Errno ,
Last_SQL_Error , and
Last_SQL_Errno in the output of
SHOW SLAVE STATUS .
(Bug#34654) See also Bug#44270. Cluster Replication: Important Note:
This release of MySQL Cluster derives in part from MySQL 5.1.29,
where the default value for the
--binlog-format option changed to
STATEMENT . That change does
not affect this or future MySQL Cluster NDB
6.x releases, where the default value for this option remains
MIXED , since MySQL Cluster Replication does
not work with the statement-based format.
(Bug#40586) Disk Data:
It is now possible to specify default locations for Disk Data
data files and undo log files, either together or separately,
using the data node configuration parameters
FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . For information
about these configuration parameters, see
Disk
Data filesystem parameters.
It is also now possible to specify a log file group, tablespace,
or both, that is created when the cluster is started, using the
InitialLogFileGroup and
InitialTablespace data node configuration
parameters. For information about these configuration
parameters, see
Disk
Data object creation parameters.
Bugs fixed: Performance:
Updates of the SYSTAB_0 system table to
obtain a unique identifier did not use transaction hints for
tables having no primary key. In such cases the NDB kernel used
a cache size of 1. This meant that each insert into a table not
having a primary key required an update of the corresponding
SYSTAB_0 entry, creating a potential
performance bottleneck.
With this fix, inserts on NDB tables without
primary keys can be under some conditions be performed up to
100% faster than previously.
(Bug#39268) Packaging:
Packages for MySQL Cluster were missing the
libndbclient.so and
libndbclient.a files.
(Bug#42278) Partitioning:
Executing ALTER TABLE ... REORGANIZE
PARTITION on an
NDBCLUSTER table having only one
partition caused mysqld to crash.
(Bug#41945) See also Bug#40389. Cluster API:
Failed operations on BLOB and
TEXT columns were not always
reported correctly to the originating SQL node. Such errors were
sometimes reported as being due to timeouts, when the actual
problem was a transporter overload due to insufficient buffer
space.
(Bug#39867, Bug#39879)
Backup IDs greater than 231 were not
handled correctly, causing negative values to be used in backup
directory names and printouts.
(Bug#43042)
When using ndbmtd, NDB kernel threads could
hang while trying to start the data nodes with
LockPagesInMainMemory set to 1.
(Bug#43021)
When using multiple management servers and starting several API
nodes (possibly including one or more SQL nodes) whose
connectstrings listed the management servers in different order,
it was possible for 2 API nodes to be assigned the same node ID.
When this happened it was possible for an API node not to get
fully connected, consequently producing a number of errors whose
cause was not easily recognizable.
(Bug#42973)
ndb_error_reporter worked correctly only with
GNU tar. (With other versions of
tar, it produced empty archives.)
(Bug#42753)
Triggers on NDBCLUSTER tables
caused such tables to become locked.
(Bug#42751) See also Bug#16229, Bug#18135.
When performing more than 32 index or tuple scans on a single
fragment, the scans could be left hanging. This caused
unnecessary timeouts, and in addition could possibly lead to a
hang of an LCP.
(Bug#42559)
A data node failure that occurred between calls to
NdbIndexScanOperation::readTuples(SF_OrderBy)
and NdbTransaction::Execute() was not
correctly handled; a subsequent call to
nextResult() caused a null pointer to be
deferenced, leading to a segfault in mysqld.
(Bug#42545)
Issuing SHOW GLOBAL STATUS LIKE 'NDB%' before
mysqld had connected to the cluster caused a
segmentation fault.
(Bug#42458)
Data node failures that occurred before all data nodes had
connected to the cluster were not handled correctly, leading to
additional data node failures.
(Bug#42422)
When a cluster backup failed with Error 1304 (Node
node_id1 : Backup request from
node_id2 failed to start), no clear
reason for the failure was provided.
As part of this fix, MySQL Cluster now retries backups in the
event of sequence errors.
(Bug#42354) See also Bug#22698.
Issuing SHOW ENGINE
NDBCLUSTER STATUS on an SQL node before the management
server had connected to the cluster caused
mysqld to crash.
(Bug#42264)
A maximum of 11 TUP scans were allowed in
parallel.
(Bug#42084)
Trying to execute an
ALTER ONLINE TABLE
... ADD COLUMN statement while inserting rows into the
table caused mysqld to crash.
(Bug#41905)
If the master node failed during a global checkpoint, it was
possible in some circumstances for the new master to use an
incorrect value for the global checkpoint index. This could
occur only when the cluster used more than one node group.
(Bug#41469)
API nodes disconnected too agressively from cluster when data
nodes were being restarted. This could sometimes lead to the API
node being unable to access the cluster at all during a rolling
restart.
(Bug#41462)
An abort path in the DBLQH kernel block
failed to release a commit acknowledgement marker. This meant
that, during node failure handling, the local query handler
could be added multiple times to the marker record which could
lead to additional node failures due an array overflow.
(Bug#41296)
During node failure handling (of a data node other than the
master), there was a chance that the master was waiting for a
GCP_NODEFINISHED signal from the failed node
after having received it from all other data nodes. If this
occurred while the failed node had a transaction that was still
being committed in the current epoch, the master node could
crash in the DBTC kernel block when
discovering that a transaction actually belonged to an epoch
which was already completed.
(Bug#41295)
If a transaction was aborted during the handling of a data node
failure, this could lead to the later handling of an API node
failure not being completed.
(Bug#41214)
Given a MySQL Cluster containing no data (that is, whose data
nodes had all been started using --initial , and
into which no data had yet been imported) and having an empty
backup directory, executing START BACKUP with
a user-specified backup ID caused the data nodes to crash.
(Bug#41031)
Issuing EXIT in the management client
sometimes caused the client to hang.
(Bug#40922)
Redo log creation was very slow on some platforms, causing MySQL
Cluster to start more slowly than necessary with some
combinations of hardware and operating system. This was due to
all write operations being synchronized to disk while creating a
redo log file. Now this synchronization occurs only after the
redo log has been created.
(Bug#40734)
Transaction failures took longer to handle than was necessary.
When a data node acting as transaction coordinator (TC) failed,
the surviving data nodes did not inform the API node initiating
the transaction of this until the failure had been processed by
all protocols. However, the API node needed only to know about
failure handling by the transaction protocol — that is, it
needed to be informed only about the TC takeover process. Now,
API nodes (including MySQL servers acting as cluster SQL nodes)
are informed as soon as the TC takeover is complete, so that it
can carry on operating more quickly.
(Bug#40697)
It was theoretically possible for stale data to be read from
NDBCLUSTER tables when the
transaction isolation level was set to
ReadCommitted .
(Bug#40543)
In some cases, NDB did not check
correctly whether tables had changed before trying to use the
query cache. This could result in a crash of the debug MySQL
server.
(Bug#40464)
Restoring a MySQL Cluster from a dump made using
mysqldump failed due to a spurious error:
Can't execute the given command because you have
active locked tables or an active transaction.
(Bug#40346)
O_DIRECT was incorrectly disabled when making
MySQL Cluster backups.
(Bug#40205)
Events logged after setting ALL CLUSTERLOG
STATISTICS=15 in the management client did not always
include the node ID of the reporting node.
(Bug#39839)
Start phase reporting was inconsistent between the management
client and the cluster log.
(Bug#39667)
The MySQL Query Cache did not function correctly with
NDBCLUSTER tables containing
TEXT columns.
(Bug#39295)
A segfault in Logger::Log caused
ndbd to hang indefinitely. This fix improves
on an earlier one for this issue, first made in MySQL Cluster
NDB 6.2.16 and MySQL Cluster NDB 6.3.17.
(Bug#39180) See also Bug#38609.
Memory leaks could occur in handling of strings used for storing
cluster metadata and providing output to users.
(Bug#38662)
In the event that a MySQL Cluster backup failed due to file
permissions issues, conflicting reports were issued in the
management client.
(Bug#34526)
A duplicate key or other error raised when inserting into an
NDBCLUSTER table caused the current
transaction to abort, after which any SQL statement other than a
ROLLBACK
failed. With this fix, the
NDBCLUSTER storage engine now
performs an implicit rollback when a transaction is aborted in
this way; it is no longer necessary to issue an explicit
ROLLBACK
statement, and the next statement that is issued automatically
begins a new transaction.
Note
It remains necessary in such cases to retry the complete
transaction, regardless of which statement caused it to be
aborted.
(Bug#32656) See also Bug#47654.
Error messages for NDBCLUSTER error
codes 1224 and 1227 were missing.
(Bug#28496) Partitioning:
A query on a user-partitioned table caused MySQL to crash, where
the query had the following characteristics:
The query's WHERE clause referenced
an indexed column that was also in the partitioning key.
The query's WHERE clause included a
value found in the partition.
The query's WHERE clause used the
< or <>
operators to compare with the indexed column's value
with a constant.
The query used an ORDER BY clause, and
the same indexed column was used in the ORDER
BY clause.
The ORDER BY clause used an explcit or
implicit ASC sort priority.
Two examples of such a query are given here, where
a represents an indexed column used in the
table's partitioning key:
SELECT * FROM table WHERE a < constant ORDER BY a;
SELECT * FROM table WHERE a <> constant ORDER BY a;
(Bug#40954) This regression was introduced by Bug#30573, Bug#33257, Bug#33555. Partitioning:
Dropping or creating an index on a partitioned table managed by
the InnoDB Plugin locked the table.
(Bug#37453) Disk Data:
It was not possible to add an in-memory column online to a table
that used a table-level or column-level STORAGE
DISK option. The same issue prevented ALTER
ONLINE TABLE ... REORGANIZE PARTITION from working on
Disk Data tables.
(Bug#42549) Disk Data:
Issuing concurrent CREATE TABLESPACE ,
ALTER TABLESPACE , CREATE LOGFILE
GROUP , or ALTER LOGFILE GROUP
statements on separate SQL nodes caused a resource leak that led
to data node crashes when these statements were used again
later.
(Bug#40921) Disk Data:
Disk-based variable-length columns were not always handled like
their memory-based equivalents, which could potentially lead to
a crash of cluster data nodes.
(Bug#39645) Disk Data:
Creating a Disk Data tablespace with a very large extent size
caused the data nodes to fail. The issue was observed when using
extent sizes of 100 MB and larger.
(Bug#39096) Disk Data:
Creation of a tablespace data file whose size was greater than 4
GB failed silently on 32-bit platforms.
(Bug#37116) See also Bug#29186. Disk Data:
O_SYNC was incorrectly disabled on platforms
that do not support O_DIRECT . This issue was
noted on Solaris but could have affected other platforms not
having O_DIRECT capability.
(Bug#34638) Disk Data:
Trying to execute a CREATE LOGFILE
GROUP statement using a value greater than
150M for UNDO_BUFFER_SIZE
caused data nodes to crash.
As a result of this fix, the upper limit for
UNDO_BUFFER_SIZE is now
600M ; attempting to set a higher value now
fails gracefully with an error.
(Bug#34102) See also Bug#36702. Disk Data:
When attempting to create a tablespace that already existed, the
error message returned was Table or index with given
name already exists.
(Bug#32662) Disk Data:
Using a path or filename longer than 128 characters for Disk
Data undo log files and tablespace data files caused a number of
issues, including failures of CREATE
LOGFILE GROUP , ALTER LOGFILE
GROUP , CREATE
TABLESPACE , and ALTER
TABLESPACE statements, as well as crashes of
management nodes and data nodes.
With this fix, the maximum length for path and file names used
for Disk Data undo log files and tablespace data files is now
the same as the maximum for the operating system.
(Bug#31769, Bug#31770, Bug#31772) Disk Data:
Starting a cluster under load such that Disk Data tables used
most of the undo buffer could cause data node failures.
The fix for this bug also corrected an issue in the
LGMAN kernel block where the amount of free
space left in the undo buffer was miscalculated, causing buffer
overruns. This could cause records in the buffer to be
overwritten, leading to problems when restarting data nodes.
(Bug#28077) Disk Data:
Attempting to perform a system restart of the cluster where
there existed a logfile group without and undo log files caused
the data nodes to crash.
Note
While issuing a CREATE LOGFILE
GROUP statement without an ADD
UNDOFILE option fails with an error in the MySQL
server, this situation could arise if an SQL node failed
during the execution of a valid CREATE
LOGFILE GROUP statement; it is also possible to
create a logfile group without any undo log files using the
NDB API.
(Bug#17614) Cluster Replication:
Sometimes, when using the --ndb_log_orig
option, the orig_epoch and
orig_server_id columns of the
ndb_binlog_index table on the slave contained
the ID and epoch of the local server instead.
(Bug#41601) Cluster API:
Some error messages from ndb_mgmd contained
newline (\n ) characters. This could break the
MGM API protocol, which uses the newline as a line separator.
(Bug#43104) Cluster API:
When using an ordered index scan without putting all key columns
in the read mask, this invalid use of the NDB API went
undetected, which resulted in the use of uninitialized memory.
(Bug#42591) Cluster API:
The MGM API reset error codes on management server handles
before checking them. This meant that calling an MGM API
function with a null handle caused applications to crash.
(Bug#40455) Cluster API:
It was not always possible to access parent objects directly
from NdbBlob ,
NdbOperation , and
NdbScanOperation objects. To alleviate this
problem, a new getNdbOperation() method has
been added to NdbBlob and new
getNdbTransaction() methods have been added to
NdbOperation and
NdbScanOperation . In addition, a const
variant of NdbOperation::getErrorLine() is
now also available.
(Bug#40242) Cluster API:
NdbScanOperation::getBlobHandle() failed when
used with incorrect column names or numbers.
(Bug#40241) Cluster API:
The NDB API example programs included in MySQL Cluster source
distributions failed to compile.
(Bug#37491) See also Bug#40238. Cluster API:
mgmapi.h contained constructs which only
worked in C++, but not in C.
(Bug#27004)
17.7.4.4. Changes in MySQL Cluster NDB 6.2.16 (5.1.28-ndb-6.2.16) (08 October 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
ftp://ftp.mysql.com/pub/mysql/download/cluster_telco/mysql-5.1.28-ndb-6.2.16.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.28 (see Section C.1.22, “Changes in MySQL 5.1.28 (28 August 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
Heavy DDL usage caused the mysqld processes
to hang due to a timeout error (NDB
error code 266).
(Bug#39885)
Executing EXPLAIN
SELECT on an NDBCLUSTER
table could cause mysqld to crash.
(Bug#39872)
Starting the MySQL Server with the
--ndbcluster option plus an
invalid command-line option (for example, using
mysqld --ndbcluster
--foobar ) caused it to hang while shutting down the
binlog thread.
(Bug#39635)
Dropping and then re-creating a database on one SQL node caused
other SQL nodes to hang.
(Bug#39613)
Setting a low value of MaxNoOfLocalScans
(< 100) and performing a large number of (certain) scans
could cause the Transaction Coordinator to run out of scan
fragment records, and then crash. Now when this resource is
exhausted, the cluster returns Error 291 (Out of
scanfrag records in TC (increase MaxNoOfLocalScans))
instead.
(Bug#39549)
Creating a unique index on an
NDBCLUSTER table caused a memory
leak in the NDB subscription
manager (SUMA ) which could lead to mysqld
hanging, due to the fact that the resource shortage was not
reported back to the NDB kernel
correctly.
(Bug#39518) See also Bug#39450.
Unique identifiers in tables having no primary key were not
cached. This fix has been observed to increase the efficiency of
INSERT operations on such tables
by as much as 50%.
(Bug#39267)
MgmtSrvr::allocNodeId() left a mutex locked
following an Ambiguity for node if %d
error.
(Bug#39158)
An invalid path specification caused
mysql-test-run.pl to fail.
(Bug#39026)
During transactional coordinator takeover (directly after node
failure), the LQH finding an operation in the
LOG_COMMIT state sent an
LQH_TRANS_CONF signal twice, causing the TC
to fail.
(Bug#38930)
An invalid memory access caused the management server to crash
on Solaris Sparc platforms.
(Bug#38628)
A segfault in Logger::Log caused
ndbd to hang indefinitely.
(Bug#38609)
ndb_mgmd failed to start on older Linux
distributions (2.4 kernels) that did not support e-polling.
(Bug#38592)
When restarting a data node, an excessively long shutodwn
message could cause the node process to crash.
(Bug#38580)
ndb_mgmd sometimes performed unnecessary
network I/O with the client. This in combination with other
factors led to long-running threads that were attempting to
write to clients that no longer existed.
(Bug#38563)
ndb_restore failed with a floating point
exception due to a division by zero error when trying to restore
certain data files.
(Bug#38520)
A failed connection to the management server could cause a
resource leak in ndb_mgmd.
(Bug#38424)
Failure to parse configuration parameters could cause a memory
leak in the NDB log parser.
(Bug#38380)
After a forced shutdown and initial restart of the cluster, it
was possible for SQL nodes to retain .frm
files corresponding to NDBCLUSTER
tables that had been dropped, and thus to be unaware that these
tables no longer existed. In such cases, attempting to re-create
the tables using CREATE TABLE IF NOT EXISTS
could fail with a spurious Table ... doesn't
exist error.
(Bug#37921)
Renaming an NDBCLUSTER table on one
SQL node, caused a trigger on this table to be deleted on
another SQL node.
(Bug#36658)
Attempting to add a UNIQUE INDEX twice to an
NDBCLUSTER table, then deleting
rows from the table could cause the MySQL Server to crash.
(Bug#35599)
ndb_restore failed when a single table was
specified.
(Bug#33801)
GCP_COMMIT did not wait for transaction
takeover during node failure. This could cause
GCP_SAVE_REQ to be executed too early. This
could also cause (very rarely) replication to skip rows.
(Bug#30780) Cluster Replication:
In some cases, dropping a database on the master could cause
table logging to fail on the slave, or, when using a debug
build, could cause the slave mysqld to fail
completely.
(Bug#39404) Cluster Replication:
During a parallel node restart, the starting nodes could
(sometimes) incorrectly synchronize subscriptions among
themselves. Instead, this synchronization now takes place only
among nodes that have actually (completely) started.
(Bug#38767) Cluster Replication:
Data was written to the binlog with
--log-slave-updates disabled.
(Bug#37472) Cluster API:
Passing a value greater than 65535 to
NdbInterpretedCode::add_val() and
NdbInterpretedCode::sub_val() caused these
methods to have no effect.
(Bug#39536) Cluster API:
The NdbScanOperation::readTuples() method
could be called multiple times without error.
(Bug#38717) Cluster API:
Certain Multi-Range Read scans involving IS
NULL and IS NOT NULL comparisons
failed with an error in the NDB
local query handler.
(Bug#38204) Cluster API:
Problems with the public headers prevented
NDB applications from being built
with warnings turned on.
(Bug#38177) Cluster API:
Creating an NdbScanFilter object using an
NdbScanOperation object that had not yet had
its readTuples() method called resulted in a
crash when later attempting to use the
NdbScanFilter .
(Bug#37986) Cluster API:
Executing an NdbRecord interpreted delete
created with an ANYVALUE option caused the
transaction to abort.
(Bug#37672) Cluster API:
Accesing the debug version of libndbclient
via dlopen() resulted in a segmentation
fault.
(Bug#35927)
17.7.4.5. Changes in MySQL Cluster NDB 6.2.15 (5.1.23-ndb-6.2.15) (22 May 2008)
This is re-release of MySQL Cluster NDB 6.2.14 providing
binaries for supported platforms. For more information, see
Section 17.7.4.6, “Changes in MySQL Cluster NDB 6.2.14 (5.1.23-ndb-6.2.14) (05 March 2008)”.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
17.7.4.6. Changes in MySQL Cluster NDB 6.2.14 (5.1.23-ndb-6.2.14) (05 March 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Added the MaxBufferedEpochs data node
configuration parameter, which controls the maximum number of
unprocessed epochs by which a subscribing node can lag.
Subscribers which exceed this number are disconnected and forced
to reconnect.
See Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for more
information.
Replication:
Introduced the slave_exec_mode
system variable to control whether idempotent or strict mode is
used for replication conflict resolution. Idempotent mode
suppresses duplicate-key, no-key-found, and some other errors,
and is needed for circular replication, multi-master
replication, and some other complex replication setups when
using MySQL Cluster, where idempotent mode is the default.
However, strict mode is the default for storage engines other
than NDB .
(Bug#31609) Cluster Replication:
RESET MASTER now uses
TRUNCATE TABLE rather than
DELETE to clear the
mysql.ndb_binlog_index table. This improves
the performance of the statement and is less likely to leave the
table in a fragmented state.
(Bug#34356)
Formerly, when the MySQL server crashed, the generated stack
dump was numeric and required external tools to properly resolve
the names of functions. This is not very helpful to users having
a limited knowledge of debugging techniques. In addition, the
generated stack trace contained only the names of functions and
was formatted differently for each platform due to different
stack layouts.
Now it is possible to take advantage of newer versions of the
GNU C Library provide a set of functions to obtain and
manipulate stack traces from within the program. On systems that
use the ELF binary format, the stack trace contains important
information such as the shared object where the call was
generated, an offset into the function, and the actual return
address. Having the function name also makes possible the name
demangling of C++ functions.
The library generates meaningful stack traces on the following
platforms: i386, x86_64, PowerPC, IA64, Alpha, and S390. On
other platforms, a numeric stack trace is still produced, and
the use of the resolve_stack_dump utility is
still required.
(Bug#31891)
mysqltest now has mkdir
and rmdir commands for creating and removing
directories.
(Bug#31004)
Added the
Uptime_since_flush_status
status variable, which indicates the number of seconds since the
most recent FLUSH STATUS statement.
(Community contribution by Jeremy Cole)
(Bug#24822)
Bugs fixed: Performance:
InnoDB exhibited thread thrashing with more
than 50 concurrent connections under an update-intensive
workload.
(Bug#22868) Incompatible Change:
The UPDATE statement allowed
NULL to be assigned to NOT
NULL columns (the implicit default value for the
column data type was assigned). This was changed so that on
error occurs.
This change was reverted, because the original report was
determined not to be a bug: Assigning NULL to
a NOT NULL column in an
UPDATE statement should produce
an error only in strict SQL mode and set the column to the
implicit default with a warning otherwise, which was the
original behavior. See Section 10.1.4, “Data Type Default Values”, and
Bug#39265.
(Bug#33699) Important Change: Replication:
When the master crashed during an update on a transactional
table while in autocommit mode,
the slave failed. This fix causes every transaction (including
autocommit transactions) to be
recorded in the binlog as starting with a
BEGIN and
ending with a COMMIT or
ROLLBACK .
(Bug#26395) Replication: Important Note:
Network timeouts between the master and the slave could result
in corruption of the relay log. This fix rectifies a
long-standing replication issue when using unreliable networks,
including replication over wide area networks such as the
Internet. If you experience reliability issues and see many
You have an error in your SQL syntax
errors on replication slaves, we strongly recommend that you
upgrade to a MySQL version which includes this fix.
(Bug#26489) Replication:
When the Windows version of mysqlbinlog read
4.1 binlogs containing
LOAD DATA
INFILE statements, it output backslashes as path
separators, causing problems for client programs expecting
forward slashes. In such cases, it now converts
\\ to / in directory
paths.
(Bug#34355) Replication:
SHOW SLAVE STATUS failed when
slave I/O was about to terminate.
(Bug#34305) Replication:
mysqlbinlog from a 5.1 or later MySQL
distribution could not read binary logs generated by a 4.1
server when the logs contained
LOAD DATA
INFILE statements.
(Bug#34141) This regression was introduced by Bug#32407. Replication:
A CREATE USER ,
DROP USER , or
RENAME USER statement that fails
on the master, or that is a duplicate of any of these
statements, is no longer written to the binlog; previously,
either of these occurrences could cause the slave to fail.
(Bug#33862) See also Bug#29749. Replication:
mysqlbinlog failed to release all of its
memory after terminating abnormally.
(Bug#33247) Replication:
The error message generated due to lack of a default value for
an extra column was not sufficiently informative.
(Bug#32971) Replication:
When a user variable was used inside an
INSERT statement, the
corresponding binlog event was not written to the binlog
correctly.
(Bug#32580) Replication:
When using row-based replication, deletes from a table with a
foreign key constraint failed on the slave.
(Bug#32468) Replication:
SQL statements containing comments using --
syntax were not replayable by mysqlbinlog,
even though such statements replicated correctly.
(Bug#32205) Replication:
When using row-based replication from a master running MySQL
5.1.21 or earlier to a slave running 5.1.22 or later, updates of
integer columns failed on the slave with Error in
Unknown event: row application failed.
(Bug#31583) This regression was introduced by Bug#21842. Replication:
Replicating write, update, or delete events from a master
running MySQL 5.1.15 or earlier to a slave running 5.1.16 or
later caused the slave to crash.
(Bug#31581) Replication:
When using row-based replication, the slave stopped when
attempting to delete nonexistent rows from a slave table without
a primary key. In addition, no error was reported when this
occurred.
(Bug#31552) Replication:
Issuing a DROP VIEW statement
caused replication to fail if the view did not actually exist.
(Bug#30998) Replication:
Replication of LOAD
DATA INFILE could fail when
read_buffer_size was larger
than max_allowed_packet .
(Bug#30435) Replication:
Replication crashed with the NDB
storage engine when mysqld was started with
--character-set-server=ucs2 .
(Bug#29562) Replication:
Setting server_id did not
update its value for the current session.
(Bug#28908) Replication:
Some older servers wrote events to the binary log using
different numbering from what is currently used, even though the
file format number in the file is the same. Slaves running MySQL
5.1.18 and later could not read these binary logs properly.
Binary logs from these older versions now are recognized and
event numbers are mapped to the current numbering so that they
can be interpreted properly.
(Bug#27779, Bug#32434) This regression was introduced by Bug#22583. Cluster Replication:
Enabling the ndb_wait_connected
system variable caused the server to wait for a partial
connection plus an additional 3 seconds for a complete
connection to the cluster. This could lead to issues with
setting up the binary log.
(Bug#34757) Cluster API:
Closing a scan before it was executed caused the application to
segfault.
(Bug#36375) Cluster API:
Using NDB API applications from older MySQL Cluster versions
with libndbclient from newer ones caused the
cluster to fail.
(Bug#36124) Cluster API:
Scans having no bounds set were handled incorrectly.
(Bug#35876)
Use of stored functions in the WHERE clause
for SHOW OPEN TABLES caused a
server crash.
(Bug#34166)
Large unsigned integers were improperly handled for prepared
statements, resulting in truncation or conversion to negative
numbers.
(Bug#33798)
The server crashed when executing a query that had a subquery
containing an equality X=Y where Y referred to a named select
list expression from the parent select. The server crashed when
trying to use the X=Y equality for
ref -based access.
(Bug#33794)
ORDER BY ... DESC sorts could produce
misordered results.
(Bug#33697)
The server could crash when
REPEAT
or another control instruction was used in conjunction with
labels and a
LEAVE
instruction.
(Bug#33618)
SET GLOBAL myisam_max_sort_file_size=DEFAULT
set myisam_max_sort_file_size
to an incorrect value.
(Bug#33382) See also Bug#31177.
Granting the UPDATE privilege on
one column of a view caused the server to crash.
(Bug#33201)
For DECIMAL columns used with the
ROUND(X ,D )
or
TRUNCATE(X ,D )
function with a nonconstant value of
D , adding an ORDER
BY for the function result produced misordered output.
(Bug#33143) See also Bug#33402, Bug#30617.
The SHOW ENGINE
INNODB STATUS and
SHOW ENGINE INNODB
MUTEX statements incorrectly required the
SUPER privilege rather than the
PROCESS privilege.
(Bug#32710)
Tables in the mysql database that stored the
current sql_mode value as part
of stored program definitions were not updated with newer mode
values
(NO_ENGINE_SUBSTITUTION ,
PAD_CHAR_TO_FULL_LENGTH ). This
causes various problems defining stored programs if those modes
were included in the current
sql_mode value.
(Bug#32633)
ROUND(X ,D )
or
TRUNCATE(X ,D )
for nonconstant values of D could
crash the server if these functions were used in an
ORDER BY that was resolved using
filesort .
(Bug#30889)
Resetting the query cache by issuing a SET GLOBAL
query_cache_size=0 statement caused the server to
crash if it concurrently was saving a new result set to the
query cache.
(Bug#30887)
The Table_locks_waited waited
variable was not incremented in the cases that a lock had to be
waited for but the waiting thread was killed or the request was
aborted.
(Bug#30331)
The Com_create_function status variable was
not incremented properly.
(Bug#30252)
mysqld displayed the
--enable-pstack option in its
help message even if MySQL was configured without
--with-pstack .
(Bug#29836)
Views were treated as insertable even if some base table columns
with no default value were omitted from the view definition.
(This is contrary to the condition for insertability that a view
must contain all columns in the base table that do not have a
default value.)
(Bug#29477)
Previously, the parser accepted the ODBC { OJ ... LEFT
OUTER JOIN ...} syntax for writing left outer joins.
The parser now allows { OJ ... } to be used
to write other types of joins, such as INNER
JOIN or RIGHT OUTER JOIN . This
helps with compatibility with some third-party applications, but
is not official ODBC syntax.
(Bug#28317)
The parser rules for the SHOW
PROFILE statement were revised to work with older
versions of bison.
(Bug#27433)
resolveip failed to produce correct results
for host names that begin with a digit.
(Bug#27427)
mysqlcheck -A -r did not correctly identify
all tables that needed repairing.
(Bug#25347)
Warnings for deprecated syntax constructs used in stored
routines make sense to report only when the routine is being
created, but they were also being reported when the routine was
parsed for loading into the execution cache. Now they are
reported only at routine creation time.
(Bug#21801)
CREATE ... SELECT did not always set
DEFAULT column values in the new table.
(Bug#21380)
If a SELECT calls a stored
function in a transaction, and a statement within the function
fails, that statement should roll back. Furthermore, if
ROLLBACK is
executed after that, the entire transaction should be rolled
back. Before this fix, the failed statement did not roll back
when it failed (even though it might ultimately get rolled back
by a ROLLBACK
later that rolls back the entire transaction).
(Bug#12713) See also Bug#34655.
17.7.4.7. Changes in MySQL Cluster NDB 6.2.13 (5.1.23-ndb-6.2.13) (22 February 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.4.8. Changes in MySQL Cluster NDB 6.2.12 (5.1.23-ndb-6.2.12) (12 February 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
Upgrades of a cluster using while a
DataMemory setting in excess of 16 GB caused
data nodes to fail.
(Bug#34378)
Performing many SQL statements on
NDB tables while in
autocommit mode caused a memory
leak in mysqld.
(Bug#34275)
In certain rare circumstances, a race condition could occur
between an aborted insert and a delete leading a data node
crash.
(Bug#34260)
Multi-table updates using ordered indexes during handling of
node failures could cause other data nodes to fail.
(Bug#34216)
When configured with NDB support,
MySQL failed to compile using gcc 4.3 on
64bit FreeBSD systems.
(Bug#34169)
The failure of a DDL statement could sometimes lead to node
failures when attempting to execute subsequent DDL statements.
(Bug#34160)
Extremely long SELECT statements
(where the text of the statement was in excess of 50000
characters) against NDB tables
returned empty results.
(Bug#34107)
Statements executing multiple inserts performed poorly on
NDB tables having
AUTO_INCREMENT columns.
(Bug#33534)
The ndb_waiter utility polled
ndb_mgmd excessively when obtaining the
status of cluster data nodes.
(Bug#32025) See also Bug#32023.
Transaction atomicity was sometimes not preserved between reads
and inserts under high loads.
(Bug#31477)
Having tables with a great many columns could cause Cluster
backups to fail.
(Bug#30172) Cluster Replication: Disk Data:
Statements violating unique keys on Disk Data tables (such as
attempting to insert NULL into a NOT
NULL column) could cause data nodes to fail. When the
statement was executed from the binlog, this could also result
in failure of the slave cluster.
(Bug#34118) Disk Data:
Updating in-memory columns of one or more rows of Disk Data
table, followed by deletion of these rows and re-insertion of
them, caused data node failures.
(Bug#33619) Cluster Replication:
Setting
--replicate-ignore-db=mysql
caused the mysql.ndb_apply_status table not
to be replicated, breaking Cluster Replication.
(Bug#28170)
17.7.4.9. Changes in MySQL Cluster NDB 6.2.11 (5.1.23-ndb-6.2.11) (28 January 2008)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster API: Important Change:
Because NDB_LE_MemoryUsage.page_size_kb shows
memory page sizes in bytes rather than kilobytes, it has been
renamed to page_size_bytes . The name
page_size_kb is now deprecated and thus
subject to removal in a future release, although it currently
remains supported for reasons of backward compatibility. See
The Ndb_logevent_type Type, for more information
about NDB_LE_MemoryUsage .
(Bug#30271)
Bugs fixed:
High numbers of insert operations, delete operations, or both
could cause NDB error 899
(Rowid already allocated) to occur
unnecessarily.
(Bug#34033)
A periodic failure to flush the send buffer by the
NDB TCP transporter could cause a
unnecessary delay of 10 ms between operations.
(Bug#34005)
A race condition could occur (very rarely) when the release of a
GCI was followed by a data node failure.
(Bug#33793)
Some tuple scans caused the wrong memory page to be accessed,
leading to invalid results. This issue could affect both
in-memory and Disk Data tables.
(Bug#33739)
The server failed to reject properly the creation of an
NDB table having an unindexed
AUTO_INCREMENT column.
(Bug#30417)
Issuing an
INSERT ...
ON DUPLICATE KEY UPDATE concurrently with or following
a TRUNCATE TABLE statement on an
NDB table failed with
NDB error 4350
Transaction already aborted.
(Bug#29851)
The Cluster backup process could not detect when there was no
more disk space and instead continued to run until killed
manually. Now the backup fails with an appropriate error when
disk space is exhausted.
(Bug#28647)
It was possible in config.ini to define
cluster nodes having node IDs greater than the maximum allowed
value.
(Bug#28298) Cluster Replication:
ndb_restore -e restored excessively large
values to the ndb_apply_status table's
epoch column when restoring to a MySQL
Cluster version supporting Micro-GCPs from an older version that
did not support these.
A workaround when restoring to MySQL Cluster releases supporting
micro-GCPs previous to MySQL Cluster NDB 6.3.8 is to perform a
32-bit shift on the epoch column values to
reduce them to their proper size.
(Bug#33406) Cluster API:
Transactions containing inserts or reads would hang during
NdbTransaction::execute() calls made from NDB
API applications built against a MySQL Cluster version that did
not support micro-GCPs accessing a later version that supported
micro-GCPs. This issue was observed while upgrading from MySQL
Cluster NDB 6.1.23 to MySQL Cluster NDB 6.2.10 when the API
application built against the earlier version attempted to
access a data node already running the later version, even after
disabling micro-GCPs by setting
TimeBetweenEpochs equal to 0.
(Bug#33895) Cluster API:
When reading a BIT(64) value using
NdbOperation:getValue() , 12 bytes were
written to the buffer rather than the expected 8 bytes.
(Bug#33750)
17.7.4.10. Changes in MySQL Cluster NDB 6.2.10 (5.1.23-ndb-6.2.10) (19 December 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.23 (see Section C.1.27, “Changes in MySQL 5.1.23 (29 January 2008)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: Partitioning:
When partition pruning on an NDB
table resulted in an ordered index scan spanning only one
partition, any descending flag for the scan was wrongly
discarded, causing ORDER BY DESC to be
treated as ORDER BY ASC ,
MAX() to be handled incorrectly, and similar
problems.
(Bug#33061)
When all data and SQL nodes in the cluster were shut down
abnormally (that is, other than by using STOP
in the cluster management client), ndb_mgm
used excessive amounts of CPU.
(Bug#33237)
When using micro-GCPs, if a node failed while preparing for a
global checkpoint, the master node would use the wrong GCI.
(Bug#32922)
Under some conditions, performing an ALTER
TABLE on an NDBCLUSTER
table failed with a Table is full error,
even when only 25% of DataMemory was in use
and the result should have been a table using less memory (for
example, changing a VARCHAR(100) column to
VARCHAR(80) ).
(Bug#32670)
17.7.4.11. Changes in MySQL Cluster NDB 6.2.9 (5.1.22-ndb-6.2.9) (22 November 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
A local checkpoint could sometimes be started before the
previous LCP was restorable from a global checkpoint.
(Bug#32519)
High numbers of API nodes on a slow or congested network could
cause connection negotiation to time out prematurely, leading to
the following issues:
(Bug#32359)
The failure of a master node could lead to subsequent failures
in local checkpointing.
(Bug#32160)
Adding a new TINYTEXT column to
an NDB table which used
COLUMN_FORMAT = DYNAMIC , and when binary
logging was enabled, caused all cluster
mysqld processes to crash.
(Bug#30213)
After adding a new column of one of the
TEXT or
BLOB types to an
NDB table which used
COLUMN_FORMAT = DYNAMIC , it was no longer
possible to access or drop the table using SQL.
(Bug#30205)
A restart of the cluster failed when more than 1 REDO phase was
in use.
(Bug#22696) Cluster Replication: Replication:
Where a table being replicated had a
TEXT or
BLOB column, an
UPDATE on the master that did not
refer explicitly to this column in the WHERE
clause stopped the SQL thread on the slave with Error
in Write_rows event: row application failed. Got error 4288
'Blob handle for column not available' from
NDBCLUSTER.
(Bug#30674) Cluster Replication:
Under certain conditions, the slave stopped processing relay
logs. This resulted in the logs never being cleared and the
slave eventually running out of disk space.
(Bug#31958)
17.7.4.12. Changes in MySQL Cluster NDB 6.2.8 (5.1.22-ndb-6.2.8) (08 November 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
In a cluster running in diskless mode and with arbitration
disabled, the failure of a data node during an insert operation
caused other data node to fail.
(Bug#31980)
An insert or update with combined range and equality constraints
failed when run against an NDB
table with the error Got unknown error from
NDB. An example of such a statement would be
UPDATE t1 SET b = 5 WHERE a IN (7,8) OR a >=
10; .
(Bug#31874)
An error with an if statement in
sql/ha_ndbcluster.cc could potentially lead
to an infinite loop in case of failure when working with
AUTO_INCREMENT columns in
NDB tables.
(Bug#31810)
The NDB storage engine code was not
safe for strict-alias optimization in gcc
4.2.1.
(Bug#31761)
Following an upgrade, ndb_mgmd would fail
with an ArbitrationError.
(Bug#31690)
The NDB management client command
node_id REPORT
MEMORY provided no output when
node_id was the node ID of a
management or API node. Now, when this occurs, the management
client responds with Node
node_id : is not a data
node .
(Bug#29485)
Performing DELETE operations
after a data node had been shut down could lead to inconsistent
data following a restart of the node.
(Bug#26450)
UPDATE IGNORE could sometimes fail on
NDB tables due to the use of
unitialized data when checking for duplicate keys to be ignored.
(Bug#25817) Cluster Replication: Replication:
A node failure during replication could lead to buckets out of
order; now active subscribers are checked for, rather than empty
buckets.
(Bug#31701) Cluster Replication:
When the master mysqld crashed or was
restarted, no LOST_EVENTS entry was made in
the binlog.
(Bug#31484) See also Bug#21494.
17.7.4.13. Changes in MySQL Cluster NDB 6.2.7 (5.1.22-ndb-6.2.7) (10 October 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster Replication:
A new configuration parameter
TimeBetweenEpochsTimeout allows a timeout to
be set for time between epochs. For more information, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
(Bug#31276)
Additional checks were implemented to catch unsupported online
ALTER TABLE operations. Currently
it is not possible to reorder columns or to change the storage
engine used for a table via online ALTER
TABLE .
Some redundant checks made during online creation of indexes
were removed.
Bugs fixed:
It was possible in some cases for a node group to be
“lost” due to missed local checkpoints following a
system restart.
(Bug#31525)
NDB tables having names containing
nonalphanumeric characters (such as
“$ ”) were not discovered
correctly.
(Bug#31470)
A node failure during a local checkpoint could lead to a
subsequent failure of the cluster during a system restart.
(Bug#31257)
A cluster restart could sometimes fail due to an issue with
table IDs.
(Bug#30975)
Transaction timeouts were not handled well in some
circumstances, leading to excessive number of transactions being
aborted unnecessarily.
(Bug#30379)
In some cases, the cluster managment server logged entries
multiple times following a restart of mgmd.
(Bug#29565)
ndb_mgm --help did not
display any information about the -a option.
(Bug#29509)
The cluster log was formatted inconsistently and contained
extraneous newline characters.
(Bug#25064)
Online ALTER operations involving a column
whose data type has an implicit default value left behind
temporary .frm files, causing subsequent
DROP DATABASE statements to fail.
(Bug#31097)
Transactions were committed prematurely when
LOCK
TABLE and SET autocommit = 0 were
used together.
(Bug#30996)
The mysqld_safe script contained a syntax
error.
(Bug#30624)
17.7.4.14. Changes in MySQL Cluster NDB 6.2.6 (5.1.22-ndb-6.2.6) (20 September 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: Partitioning:
EXPLAIN
PARTITIONS reported partition usage by queries on
NDB tables according to the
standard MySQL hash function than the hash function used in the
NDB storage engine.
(Bug#29550)
When an NDB event was left behind
but the corresponding table was later recreated and received a
new table ID, the event could not be dropped.
(Bug#30877)
Attempting to restore a backup made on a cluster host using one
endian to a machine using the other endian could cause the
cluster to fail.
(Bug#29674)
The description of the --print option provided
in the output from ndb_restore --help
was incorrect.
(Bug#27683)
Restoring a backup made on a cluster host using one endian to a
machine using the other endian failed for
BLOB and
DATETIME columns.
(Bug#27543, Bug#30024)
An insufficiently descriptive and potentially misleading Error
4006 (Connect failure - out of connection
objects...) was produced when either of the
following two conditions occurred:
There were no more transaction records in the transaction
coordinator
An NDB object in the NDB API
was initialized with insufficient parallelism
Separate error messages are now generated for each of these two
cases.
(Bug#11313)
For micro-GCPs, the assignment of “fake” CGI events
no longer cause buckets to be sent out of order. Now, when
assigning a GCI to a non-GCI event (that is, creating a
pseudo-GCI or “fake” CGI), the GCI that is to
arrive is always initiated, even if no known GCI exists, which
could occur in the event of a node failure.
(Bug#30884)
17.7.4.15. Changes in MySQL Cluster NDB 6.2.5 (5.1.22-ndb-6.2.5) (06 September 2007 General Availability)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.22 (see Section C.1.28, “Changes in MySQL 5.1.22 (24 September 2007 Release Candidate)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
The following improvements have been made in the
ndb_size.pl utility:
The script can now be used with multiple databases; lists of
databases and tables can also be excluded from analysis.
Schema name information has been added to index table
calculations.
The database name is now an optional parameter, the
exclusion of which causes all databases to be examined.
If selecting from INFORMATION_SCHEMA
fails, the script now attempts to fall back to
SHOW TABLES .
A --real_table_name option has been added;
this designates a table to handle unique index size
calculations.
The report title has been amended to cover cases where more
than one database is being analyzed.
Support for a --socket option was also added.
For more information, see
Section 17.4.21, “ndb_size.pl — NDBCLUSTER Size Requirement Estimator”.
(Bug#28683, Bug#28253)
Online ADD COLUMN , ADD
INDEX , and DROP INDEX
operations can now be performed explicitly for
NDB tables, as well as online
renaming of tables and columns for
NDB and MyISAM
tables — that is, without copying or locking of the
affected tables — using ALTER ONLINE
TABLE .
Indexes can also be created and dropped online using
CREATE INDEX and
DROP INDEX , respectively, using
the ONLINE keyword.
You can force operations that would otherwise be performed
online to be done offline using the OFFLINE
keyword.
See Section 12.1.7, “ALTER TABLE Syntax”,
Section 12.1.13, “CREATE INDEX Syntax”, and
Section 12.1.24, “DROP INDEX Syntax”, for more information.
It is now possible to control whether fixed-width or
variable-width storage is used for a given column of an
NDB table by means of the
COLUMN_FORMAT specifier as part of the
column's definition in a CREATE
TABLE or ALTER TABLE
statement.
It is also possible to control whether a given column of an
NDB table is stored in memory or on
disk, using the STORAGE specifier as part of
the column's definition in a CREATE
TABLE or ALTER TABLE
statement.
For permitted values and other information about
COLUMN_FORMAT and STORAGE ,
see Section 12.1.17, “CREATE TABLE Syntax”.
A new cluster management server startup option
--bind-address makes it possible
to restrict management client connections to
ndb_mgmd to a single host and port. For more
information, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
Cluster Replication:
The protocol for handling global checkpoints has been changed.
It is now possible to control how often the GCI number is
updated, and how often global checkpoints are written to disk,
using the TimeBetweenEpochs configuration
parameter. This improves the reliability and performance of
MySQL Cluster Replication.
GCPs handled using the new protocol are sometimes referred to as
“micro-GCPs”.
For more information, see
TimeBetweenEpochs
.
Bugs fixed:
When handling BLOB columns, the
addition of read locks to the lock queue was not handled
correctly.
(Bug#30764)
Discovery of NDB tables did not
work correctly with INFORMATION_SCHEMA .
(Bug#30667)
A file system close operation could fail during a node or system
restart.
(Bug#30646)
Using the --ndb-cluster-connection-pool option
for mysqld caused DDL statements to be
executed twice.
(Bug#30598)
When creating an NDB table with a column that
has COLUMN_FORMAT = DYNAMIC , but the table
tiself uses ROW_FORMAT=FIXED , the table is
considered dynamic, but any columns for which the row format is
unspecified default to FIXED . Now in such
cases the server issues the warning Row format FIXED
incompatible with dynamic attribute
column_name .
(Bug#30276)
ndb_size.pl failed on tables with
FLOAT columns whose definitions
included commas (for example, FLOAT(6,2) ).
(Bug#29228)
Reads on BLOB columns were not
locked when they needed to be to guarantee consistency.
(Bug#29102) See also Bug#31482.
A query using joins between several large tables and requiring
unique index lookups failed to complete, eventually returning
Uknown Error after a very long period of
time. This occurred due to inadequate handling of instances
where the Transaction Coordinator ran out of
TransactionBufferMemory , when the cluster
should have returned NDB error code 4012 (Request
ndbd time-out).
(Bug#28804)
An attempt to perform a SELECT ... FROM
INFORMATION_SCHEMA.TABLES whose result included
information about NDB tables for
which the user had no privileges crashed the MySQL Server on
which the query was performed.
(Bug#26793) Cluster Replication:
Cluster replication did not handle large
VARCHAR columns correctly.
(Bug#29904) Cluster Replication:
An issue with the mysql.ndb_apply_status
table could cause NDB schema
autodiscovery to fail in certain rare circumstances.
(Bug#20872) Cluster API:
A call to CHECK_TIMEDOUT_RET() in
mgmapi.cpp should have been a call to
DBUG_CHECK_TIMEDOUT_RET() .
(Bug#30681)
17.7.4.16. Changes in MySQL Cluster NDB 6.2.4 (5.1.19-ndb-6.2.4) (04 July 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.19 (see Section C.1.31, “Changes in MySQL 5.1.19 (25 May 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
When restarting a data node, queries could hang during that
node's start phase 5, and continue only after the node had
entered phase 6.
(Bug#29364)
Replica redo logs were inconsistently handled during a system
restart.
(Bug#29354) Disk Data:
Performing Disk Data schema operations during a node restart
could cause forced shutdowns of other data nodes.
(Bug#29501) Disk Data:
Disk data meta-information that existed in
ndbd might not be visible to
mysqld.
(Bug#28720) Disk Data:
The number of free extents was incorrectly reported for some
tablespaces.
(Bug#28642)
Batching of transactions was not handled correctly in some
cases.
(Bug#29525)
17.7.4.17. Changes in MySQL Cluster NDB 6.2.3 (5.1.19-ndb-6.2.3) (02 July 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.18 (see Section C.1.32, “Changes in MySQL 5.1.18 (08 May 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
When restarting a data node, queries could hang during that
node's start phase 5, and continue only after the node had
entered phase 6.
(Bug#29364)
Replica redo logs were inconsistently handled during a system
restart.
(Bug#29354) Disk Data:
Performing Disk Data schema operations during a node restart
could cause forced shutdowns of other data nodes.
(Bug#29501) Disk Data:
Disk data meta-information that existed in
ndbd might not be visible to
mysqld.
(Bug#28720) Disk Data:
The number of free extents was incorrectly reported for some
tablespaces.
(Bug#28642)
Batching of transactions was not handled correctly in some
cases.
(Bug#29525)
17.7.4.18. Changes in MySQL Cluster NDB 6.2.2 (5.1.18-ndb-6.2.2) (07 May 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.2 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.18 (see Section C.1.32, “Changes in MySQL 5.1.18 (08 May 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
New cluster management client DUMP commands
were added to aid in tracking transactions, scan operations, and
locks. See DUMP 2350 ,
DUMP 2352 , and
DUMP 2550 , for more
information.
Added the mysqld option
--ndb-cluster-connection-pool that allows a
single MySQL server to use multiple connections to the cluster.
This allows for scaling out using multiple MySQL clients per SQL
node instead of or in addition to using multiple SQL nodes with
the cluster.
For more information about this option, see
Section 17.3.4, “MySQL Server Options and Variables for MySQL Cluster”.
17.7.4.19. Changes in MySQL Cluster NDB 6.2.1 (5.1.18-ndb-6.2.1) (30 April 2007)
This is a Beta development release, fixing recently discovered
bugs in the previous MySQL Cluster NDB 6.2 release.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.2 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.18 (see Section C.1.32, “Changes in MySQL 5.1.18 (08 May 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
Multiple operations involving deletes followed by reads were not
handled correctly.
Note
This issue could also affect MySQL Cluster Replication.
(Bug#28276) Cluster API:
Using NdbBlob::writeData() to write data in
the middle of an existing blob value (that is, updating the
value) could overwrite some data past the end of the data to be
changed.
(Bug#27018)
Incorrect handling of fragmentation in a node takeover during a
restart could cause stale data to be copied to the starting
node, leading eventually to failure of the node.
(Bug#27434)
An incorrect assertion was made when sending a
TCKEYFAILREF or TCKEYCONF
message to a failed data node.
(Bug#26814)
17.7.4.20. Changes in MySQL Cluster NDB 6.2.0 (5.1.16-ndb-6.2.0) (03 March 2007 Beta)
This is the first MySQL Cluster NDB 6.2 development release,
based on version 6.2 of the
NDBCLUSTER storage engine.
Obtaining MySQL Cluster NDB 6.2.
You can download the latest MySQL Cluster NDB 6.2 source code
and binaries for supported platforms from
http://dev.mysql.com/downloads/select.php?id=14.
This Beta release incorporates bugfixes and changes made in
previous MySQL Cluster releases, as well as all bugfixes and
feature changes which were added in mainline MySQL 5.1 through
MySQL 5.1.16 (see Section C.1.34, “Changes in MySQL 5.1.16 (26 February 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
An ndb_wait_connected system
variable has been added for mysqld. It causes
mysqld wait a specified amount of time to be
connected to the cluster before accepting client connections.
For more information, see
Section 17.3.4.3, “MySQL Cluster System Variables”.
Cluster API:
The Ndb::startTransaction() method now
provides an alternative interface for starting a transaction.
See Ndb::startTransaction() , for more
information.
Cluster API:
Methods were added to the
Ndb_cluster_connection class to faciliate
iterating over existing NDB
objects. See
ndb_cluster_connection::get_next_ndb_object() ,
for more information.
It is now possible to disable arbitration by setting
ArbitrationRank equal to 0
on all nodes.
A new TcpBind_INADDR_ANY configuration
parameter allows data nodes node to bind
INADDR_ANY instead of a host name or IP
address in the config.ini file.
Memory allocation has been improved on 32-bit architectures that
enables using close to 3GB for DataMemory and
IndexMemory combined.
17.7.5. Changes in MySQL Cluster NDB 6.1
This section contains change history information for MySQL Cluster
releases based on version 6.1 of the
NDBCLUSTER storage engine.
For an overview of features that were added added in MySQL Cluster
NDB 6.1, see
Section 17.1.4.2, “MySQL Cluster Development in MySQL Cluster NDB 6.1”.
Note
MySQL Cluster NDB 6.1 is no longer being developed or
maintained, and the information presented in the next few
sections should be considered to be of historical interest only.
If you are using MySQL Cluster NDB 6.1, you should upgrade as
soon as possible to the most recent version of MySQL Cluster NDB
6.2 or later MySQL Cluster release series.
17.7.5.1. Changes in MySQL Cluster NDB 6.1.23 (5.1.15-ndb-6.1.23) (20 November 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
The NDB storage engine code was not
safe for strict-alias optimization in gcc
4.2.1.
(Bug#31761) Cluster Replication:
Under certain conditions, the slave stopped processing relay
logs. This resulted in the logs never being cleared and the
slave eventually running out of disk space.
(Bug#31958)
17.7.5.2. Changes in MySQL Cluster NDB 6.1.22 (5.1.15-ndb-6.1.22) (19 October 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
It was possible in some cases for a node group to be
“lost” due to missed local checkpoints following a
system restart.
(Bug#31525) Cluster Replication: Replication:
A node failure during replication could lead to buckets out of
order; now active subscribers are checked for, rather than empty
buckets.
(Bug#31701)
17.7.5.3. Changes in MySQL Cluster NDB 6.1.21 (5.1.15-ndb-6.1.21) (01 October 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
A node failure during a local checkpoint could lead to a
subsequent failure of the cluster during a system restart.
(Bug#31257)
A cluster restart could sometimes fail due to an issue with
table IDs.
(Bug#30975)
17.7.5.4. Changes in MySQL Cluster NDB 6.1.20 (5.1.15-ndb-6.1.20) (14 September 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.5.5. Changes in MySQL Cluster NDB 6.1.19 (5.1.15-ndb-6.1.19) (01 August 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
Whenever a TCP send buffer is over 80% full, temporary error
1218 (Send Buffers overloaded in NDB
kernel) is now returned. See
SendBufferMemory
for more information.
An INFO event is now sent if the time between
global checkpoints is excessive, or if DUMP
7901 is issued in the management client.
17.7.5.6. Changes in MySQL Cluster NDB 6.1.18 (5.1.15-ndb-6.1.18) (Not released)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
When restarting a data node, queries could hang during that
node's start phase 5, and continue only after the node had
entered phase 6.
(Bug#29364) Replication:
Storage engine error conditions in row-based replication were
not correctly reported to the user.
(Bug#29570) Disk Data:
Disk data meta-information that existed in
ndbd might not be visible to
mysqld.
(Bug#28720) Disk Data:
The number of free extents was incorrectly reported for some
tablespaces.
(Bug#28642)
17.7.5.7. Changes in MySQL Cluster NDB 6.1.17 (5.1.15-ndb-6.1.17) (03 July 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: 17.7.5.8. Changes in MySQL Cluster NDB 6.1.16 (5.1.15-ndb-6.1.16) (29 June 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
When a node failed to respond to a COPY_GCI
signal as part of a global checkpoint, the master node was
killed instead of the node that actually failed.
(Bug#29331)
An invalid comparison made during REDO
validation that could lead to an Error while reading
REDO log condition.
(Bug#29118)
The wrong data pages were sometimes invalidated following a
global checkpoint.
(Bug#29067)
If at least 2 files were involved in REDO
invalidation, then file 0 of page 0 was not updated and so
pointed to an invalid part of the redo log.
(Bug#29057) Disk Data:
When dropping a page, the stack's bottom entry could sometime be
left “cold” rather than “hot”,
violating the rules for stack pruning.
(Bug#29176)
17.7.5.9. Changes in MySQL Cluster NDB 6.1.15 (5.1.15-ndb-6.1.15) (20 June 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.5.10. Changes in MySQL Cluster NDB 6.1.14 (5.1.15-ndb-6.1.14) (19 June 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
In the event that two data nodes in the same node group and
participating in a GCP crashed before they had written their
respective P0.sysfile files,
QMGR could refuse to start, issuing an
invalid Insufficient nodes for restart
error instead.
(Bug#29167)
17.7.5.11. Changes in MySQL Cluster NDB 6.1.13 (5.1.15-ndb-6.1.13) (15 June 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed: 17.7.5.12. Changes in MySQL Cluster NDB 6.1.12 (5.1.15-ndb-6.1.12) (13 June 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
New cluster management client DUMP commands
were added to aid in tracking transactions, scan operations, and
locks. See DUMP 2350 ,
DUMP 2352 , and
DUMP 2550 .
Backup dump output was extended to provide more information.
Bugs fixed: 17.7.5.13. Changes in MySQL Cluster NDB 6.1.11 (5.1.15-ndb-6.1.11) (06 June 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Important Change:
The TimeBetweenWatchdogCheckInitial
configuration parameter was added to allow setting of a separate
watchdog timeout for memory allocation during startup of the
data nodes. See Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”,
for more information.
(Bug#28899)
A new configuration parameter ODirect causes
NDB to attempt using
O_DIRECT writes for LCP, backups, and redo
logs, often lowering CPU usage.
It is now possible to set the size of redo log files (fragment
log files) using the FragmentLogFileSize
configuration parameter.
Bugs fixed:
Having large amounts of memory locked caused swapping to disk.
(Bug#28751)
LCP files were not removed following an initial system restart.
(Bug#28726) Disk Data:
Repeated INSERT and
DELETE operations on a Disk Data
table having one or more large
VARCHAR columns could cause data
nodes to fail.
(Bug#20612)
17.7.5.14. Changes in MySQL Cluster NDB 6.1.10 (5.1.15-ndb-6.1.10) (30 May 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
A new times printout was added in the
ndbd watchdog thread.
Some unneeded printouts in the ndbd out file
were removed.
The names of some log and other files were changed to avoid
issues with the tar command's 99-character
file name limit.
Bugs fixed:
A regression in the heartbeat monitoring code could lead to node
failure under high load. This issue affected MySQL 5.1.19 and
MySQL Cluster NDB 6.1.10 only.
(Bug#28783)
A corrupt schema file could cause a File already
open error.
(Bug#28770)
Setting InitialNoOpenFiles equal to
MaxNoOfOpenFiles caused an error. This was
due to the fact that the actual value of
MaxNoOfOpenFiles as used by the cluster was
offset by 1 from the value set in
config.ini .
(Bug#28749)
A race condition could result when nonmaster nodes (in addition
to the master node) tried to update active status due to a local
checkpoint (that is, between NODE_FAILREP and
COPY_GCIREQ events). Now only the master
updates the active status.
(Bug#28717)
A fast global checkpoint under high load with high usage of the
redo buffer caused data nodes to fail.
(Bug#28653) Disk Data:
When loading data into a cluster following a version upgrade,
the data nodes could forcibly shut down due to page and buffer
management failures (that is, ndbrequire
failures in PGMAN ).
(Bug#28525)
17.7.5.15. Changes in MySQL Cluster NDB 6.1.9 (5.1.15-ndb-6.1.9) (24 May 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
When an API node sent more than 1024 signals in a single batch,
NDB would process only the first
1024 of these, and then hang.
(Bug#28443) Disk Data:
The cluster backup process scanned in ACC
index order, which had bad effects for disk data.
(Bug#28593)
17.7.5.16. Changes in MySQL Cluster NDB 6.1.8 (5.1.15-ndb-6.1.8) (05 May 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed:
Local checkpoint files relating to dropped
NDB tables were not removed.
(Bug#28348)
Repeated insertion of data generated by
mysqldump into
NDB tables could eventually lead to
failure of the cluster.
(Bug#27437) Disk Data:
Extremely large inserts into Disk Data tables could lead to data
node failure in some circumstances.
(Bug#27942) Cluster API:
In a multi-operation transaction, a delete operation followed by
the insertion of an implicit NULL failed to
overwrite an existing value.
(Bug#20535)
Setting MaxNoOfTables very low and relative
to DataMemory caused Out of memory
in Ndb Kernel errors when inserting relatively small
amounts of data into NDB tables.
(Bug#24173)
17.7.5.17. Changes in MySQL Cluster NDB 6.1.7 (5.1.15-ndb-6.1.7) (05 May 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster Replication: Incompatible Change:
The schema for the ndb_apply_status table in
the mysql system database has changed. When
upgrading to this release from a previous MySQL Cluster NDB 6.x
or mainline MySQL 5.1 release, you must drop the
mysql.ndb_apply_status table, then restart
the server in order for the table to be re-created with the new
schema.
See Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
additional information.
Bugs fixed:
The cluster waited 30 seconds instead of 30 milliseconds before
reading table statistics.
(Bug#28093)
Under certain rare circumstances, ndbd could
get caught in an infinite loop when one transaction took a read
lock and then a second transaction attempted to obtain a write
lock on the same tuple in the lock queue.
(Bug#28073)
Under some circumstances, a node restart could fail to update
the Global Checkpoint Index (GCI).
(Bug#28023)
An INSERT followed by a delete
DELETE on the same
NDB table caused a memory leak.
(Bug#27756) This regression was introduced by Bug#20612.
Under certain rare circumstances performing a
DROP TABLE or
TRUNCATE TABLE on an
NDB table could cause a node
failure or forced cluster shutdown.
(Bug#27581)
Memory usage of a mysqld process grew even
while idle.
(Bug#27560)
Performing a delete followed by an insert during a local
checkpoint could cause a Rowid already
allocated error.
(Bug#27205) Cluster Replication: Disk Data:
An issue with replication of Disk Data tables could in some
cases lead to node failure.
(Bug#28161) Disk Data:
Changes to a Disk Data table made as part of a transaction could
not be seen by the client performing the changes until the
transaction had been committed.
(Bug#27757) Disk Data:
When restarting a data node following the creation of a large
number of Disk Data objects (approximately 200 such objects),
the cluster could not assign a node ID to the restarting node.
(Bug#25741) Disk Data:
Changing a column specification or issuing a
TRUNCATE TABLE statement on a
Disk Data table caused the table to become an in-memory table.
This fix supersedes an incomplete fix that was made for this
issue in MySQL 5.1.15.
(Bug#24667, Bug#25296) Cluster Replication:
Some queries that updated multiple tables were not backed up
correctly.
(Bug#27748) Cluster Replication:
It was possible for API nodes to begin interacting with the
cluster subscription manager before they were fully connected to
the cluster.
(Bug#27728) Cluster Replication:
Under very high loads, checkpoints could be read or written with
checkpoint indexes out of order.
(Bug#27651) Cluster API:
An issue with the way in which the
NdbDictionary::Dictionary::listEvents()
method freed resources could sometimes lead to memory
corruption.
(Bug#27663)
mysqldump could not dump log tables.
(Bug#26121)
The --with-readline option for
configure did not work for commercial source
packages, but no error message was printed to that effect. Now a
message is printed.
(Bug#25530)
17.7.5.18. Changes in MySQL Cluster NDB 6.1.6 (5.1.15-ndb-6.1.6) (Not released)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster Replication: Incompatible Change:
The schema for the ndb_apply_status table in
the mysql system database has changed. When
upgrading to this release from a previous MySQL Cluster NDB 6.x
or mainline MySQL 5.1 release, you must drop the
mysql.ndb_apply_status table, then restart
the server in order for the table to be re-created with the new
schema.
See Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
additional information.
Bugs fixed:
A data node failing while another data node was restarting could
leave the cluster in an inconsistent state. In certain rare
cases, this could lead to a race condition and the eventual
forced shutdown of the cluster.
(Bug#27466)
It was not possible to set
LockPagesInMainMemory equal to
0 .
(Bug#27291)
A race condition could sometimes occur if the node acting as
master failed while node IDs were still being allocated during
startup.
(Bug#27286)
When a data node was taking over as the master node, a race
condition could sometimes occur as the node was assuming
responsibility for handling of global checkpoints.
(Bug#27283)
mysqld could crash shortly after a data node
failure following certain DML operations.
(Bug#27169)
The same failed request from an API node could be handled by the
cluster multiple times, resulting in reduced performance.
(Bug#27087)
The failure of a data node while restarting could cause other
data nodes to hang or crash.
(Bug#27003)
mysqld processes would sometimes crash under
high load.
Note
This fix improves on and replaces a fix for this bug that was
made in MySQL Cluster NDB 6.1.5.
(Bug#26825) Disk Data:
DROP INDEX on a Disk Data table
did not always move data from memory into the tablespace.
(Bug#25877) Cluster Replication:
Trying to replicate a large number of frequent updates with a
relatively small relay log
(max-relay-log-size set to 1M or less) could
cause the slave to crash.
(Bug#27529) Cluster API:
An issue with the way in which the
NdbDictionary::Dictionary::listEvents()
method freed resources could sometimes lead to memory
corruption.
(Bug#27663) Cluster API:
A delete operation using a scan followed by an insert using a
scan could cause a data node to fail.
(Bug#27203)
17.7.5.19. Changes in MySQL Cluster NDB 6.1.5 (5.1.15-ndb-6.1.5) (15 March 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Cluster Replication: Incompatible Change:
The schema for the ndb_apply_status table in
the mysql system database has changed. When
upgrading to this release from a previous MySQL Cluster NDB 6.x
or mainline MySQL 5.1 release, you must drop the
mysql.ndb_apply_status table, then restart
the server in order for the table to be re-created with the new
schema.
See Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
additional information.
Bugs fixed:
Creating a table on one SQL node while in single user mode
caused other SQL nodes to crash.
(Bug#26997)
mysqld processes would sometimes crash under
high load.
Note
This fix was reverted in MySQL Cluster NDB 6.1.6.
(Bug#26825)
An infinite loop in an internal logging function could cause
trace logs to fill up with Unknown Signal
type error messages and thus grow to unreasonable
sizes.
(Bug#26720) Disk Data:
When creating a log file group, setting
INITIAL_SIZE to less than
UNDO_BUFFER_SIZE caused data nodes to crash.
(Bug#25743)
17.7.5.20. Changes in MySQL Cluster NDB 6.1.4 (5.1.15-ndb-6.1.4) (09 March 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
An ndb_wait_connected system
variable has been added for mysqld. It causes
mysqld wait a specified amount of time to be
connected to the cluster before accepting client connections.
For more information, see
Section 17.3.4.3, “MySQL Cluster System Variables”.
Cluster API:
It is now possible to specify the transaction coordinator when
starting a transaction. See
Ndb::startTransaction() , for more
information.
Cluster API:
It is now possible to iterate over all existing
NDB objects using three new methods
of the Ndb_cluster_connection class:
lock_ndb_objects()
get_next_ndb_object()
unlock_ndb_objects()
For more information about these methods and their use, see
ndb_cluster_connection::get_next_ndb_object() ,
in the MySQL Cluster API Guide.
Data node memory allocation has been improved. On 32-bit
platforms, it should now be possible to use close to 3GB RAM for
IndexMemory and DataMemory
combined.
Bugs fixed:
Using only the --print_data option (and no
other options) with ndb_restore caused
ndb_restore to fail.
(Bug#26741) This regression was introduced by Bug#14612.
An inadvertent use of unaligned data caused
ndb_restore to fail on some 64-bit platforms,
including Sparc and Itanium-2.
(Bug#26739)
Assigning a node ID greater than 63 to an SQL node caused an out
of bounds error in mysqld. It should now be
possible to assign to SQL nodes node IDs up to 255.
(Bug#26663)
17.7.5.21. Changes in MySQL Cluster NDB 6.1.3 (5.1.15-ndb-6.1.3) (25 February 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
An invalid pointer was returned following a
FSCLOSECONF signal when accessing the REDO
logs during a node restart or system restart.
(Bug#26515)
The InvalidUndoBufferSize error used the
same error code (763) as the
IncompatibleVersions error.
InvalidUndoBufferSize now uses its own
error code (779).
(Bug#26490)
The failure of a data node when restarting it with
--initial could lead to failures of subsequent
data node restarts.
(Bug#26481)
Takeover for local checkpointing due to multiple failures of
master nodes was sometimes incorrectly handled.
(Bug#26457)
The LockPagesInMainMemory parameter was not
read until after distributed communication had already started
between cluster nodes. When the value of this parameter was
1 , this could sometimes result in data node
failure due to missed heartbeats.
(Bug#26454)
Under some circumstances, following the restart of a management
node, all data nodes would connect to it normally, but some of
them subsequently failed to log any events to the management
node.
(Bug#26293)
No appropriate error message was provided when there was
insufficient REDO log file space for the cluster to start.
(Bug#25801)
A memory allocation failure in SUMA (the
cluster Subscription Manager) could cause the cluster to crash.
(Bug#25239)
The message Error 0 in readAutoIncrementValue(): no
Error was written to the error log whenever
SHOW TABLE STATUS was performed
on a Cluster table that did not have an
AUTO_INCREMENT column.
Note
This improves on and supersedes an earlier fix that was made
for this issue in MySQL 5.1.12.
(Bug#21033) Disk Data:
A memory overflow could occur with tables having a large amount
of data stored on disk, or with queries using a very high degree
of parallelism on Disk Data tables.
(Bug#26514) Disk Data:
Use of a tablespace whose INITIAL_SIZE was
greater than 1 GB could cause the cluster to crash.
(Bug#26487)
17.7.5.22. Changes in MySQL Cluster NDB 6.1.2 (5.1.15-ndb-6.1.2) (07 February 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
previous MySQL Cluster NDB 6.1 releases, as well as all bugfixes
and feature changes which were added in mainline MySQL 5.1
through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Bugs fixed: 17.7.5.23. Changes in MySQL Cluster NDB 6.1.1 (5.1.15-ndb-6.1.1) (01 February 2007)
This is a bugfix release, fixing recently discovered bugs in the
previous MySQL Cluster NDB 6.1 release.
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
This Beta release incorporates all bugfixes and changes made in
the previous MySQL Cluster NDB 6.1 release, as well as all
bugfixes and feature changes which were added in mainline MySQL
5.1 through MySQL 5.1.15 (see Section C.1.35, “Changes in MySQL 5.1.15 (25 January 2007)”).
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed: Bugs fixed:
A memory leak could cause problems during a node or cluster
shutdown or failure.
(Bug#25997) Cluster API: Disk Data:
A delete and a read performed in the same operation could cause
one or more data nodes to crash. This could occur when the
operation affected more than 5 columns concurrently, or when one
or more of the columns was of the
VARCHAR type and was stored on
disk.
(Bug#25794)
An element could sometimes be inserted twice into the hash
table, causing a data node to crash.
(Bug#25286)
17.7.5.24. Changes in MySQL Cluster NDB 6.1.0 (5.1.14-ndb-6.1.0) (20 December 2006)
This is the first MySQL Cluster NDB 6.1 release, incorporating
new features and bugfixes made for the
NDBCLUSTER storage engine made
since branching from MySQL 5.1.14 standard (see
Section C.1.36, “Changes in MySQL 5.1.14 (05 December 2006)”).
MySQL Cluster NDB 6.1 no longer in development.
MySQL Cluster NDB 6.1 (formerly known as “MySQL Cluster
Carrier Grade Edition 6.1.x”) is no longer being
developed or maintained; if you are using a MySQL Cluster NDB
6.1 release, you should consider upgrading to MySQL Cluster
NDB 6.2 or 6.3.
Note
Please refer to our bug database at
http://bugs.mysql.com/ for more details about
the individual bugs fixed in this version.
Functionality added or changed:
A new configuration parameter
MemReportFrequency allows for additional
control of data node memory usage. Previously, only warnings at
predetermined percentages of memory allocation were given;
setting this parameter allows for that behavior to be
overridden. For more information, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
Bugs fixed:
When a data node was shut down using the management client
STOP command, a connection event
(NDB_LE_Connected ) was logged instead of a
disconnection event (NDB_LE_Disconnected ).
(Bug#22773)
SELECT statements with a
BLOB or
TEXT column in the selected
column list and a WHERE condition including a
primary key lookup on a VARCHAR
primary key produced empty result sets.
(Bug#19956) Disk Data:
MEDIUMTEXT columns of Disk Data
tables were stored in memory rather than on disk, even if the
columns were not indexed.
(Bug#25001) Disk Data:
Performing a node restart with a newly dropped Disk Data table
could lead to failure of the node during the restart.
(Bug#24917) Disk Data:
When restoring from backup a cluster containing any Disk Data
tables with hidden primary keys, a node failure resulted which
could lead to a crash of the cluster.
(Bug#24166) Disk Data:
Repeated CREATE , DROP , or
TRUNCATE TABLE in various
combinations with system restarts between these operations could
lead to the eventual failure of a system restart.
(Bug#21948) Disk Data:
Extents that should have been available for re-use following a
DROP TABLE operation were not
actually made available again until after the cluster had
performed a local checkpoint.
(Bug#17605) Cluster API:
Invoking the NdbTransaction::execute() method
using execution type Commit and abort option
AO_IgnoreError could lead to a crash of the
transaction coordinator (DBTC ).
(Bug#25090) Cluster API:
A unique index lookup on a nonexistent tuple could lead to a
data node timeout (error 4012).
(Bug#25059) Cluster API:
When using the NdbTransaction::execute()
method, a very long timeout (greater than 5 minutes) could
result if the last data node being polled was disconnected from
the cluster.
(Bug#24949) Cluster API:
Due to an error in the computation of table fragment arrays,
some transactions were not executed from the correct starting
point.
(Bug#24914)
Under certain rare circumstances, local checkpoints were not
performed properly, leading to an inability to restart one or
more data nodes.
(Bug#24664)
17.7.6. Release Series Changelogs — MySQL Cluster NDB 6.X and 7.X
This section contains unified changelog information for each MySQL
Cluster release series (NDB 6.1, NDB 6.2, NDB 6.3, and NDB 7.0).
For changelogs covering individual MySQL Cluster NDB 6.X and MySQL
Cluster NDB 7.X releases, see
Section 17.7, “Changes in MySQL Cluster NDB 6.X and 7.X”.
For general information about features added in MySQL Cluster NDB
6.X and 7.X, see Section 17.1.4, “MySQL Cluster Development History”.
An overview of features added in MySQL 5.1 not specific to MySQL
Cluster can be found here: Section 1.5, “What Is New in MySQL 5.1”. For a
complete list of all bugfixes and features changes made in MySQL
5.1 that are not specific to MySQL Cluster, see
Section C.1, “Changes in Release 5.1.x (Production)”.
17.7.6.1. Changes in the MySQL Cluster NDB 7.0 Series
This section contains unified change history highlights for all
MySQL Cluster releases based on version 7.0 of the
NDBCLUSTER storage engine through
MySQL Cluster NDB 7.0.12. Included are all
changelog entries in the categories MySQL
Cluster, Disk Data, and
Cluster API.
Early MySQL Cluster NDB 7.0 releases tagged “NDB
6.4.x” are also included in this listing.
For an overview of features that were added in MySQL Cluster NDB
7.0, see
Section 17.1.4.5, “MySQL Cluster Development in MySQL Cluster NDB 7.0”.
Changes in MySQL Cluster NDB 7.0.10 (5.1.39-ndb-7.0.10) Functionality added or changed:
Added the ndb_mgmd
--nowait-nodes option, which
allows a cluster that is configured to use multiple management
servers to be started using fewer than the number configured.
This is most likely to be useful when a cluster is configured
with two management servers and you wish to start the cluster
using only one of them.
See Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”, for more
information.
(Bug#48669)
This enhanced functionality is supported for upgrades from MySQL
Cluster NDB 6.3 when the NDB engine
version is 6.3.29 or later.
(Bug#48528, Bug#49163)
The output from ndb_config
--configinfo
--xml now indicates, for each
configuration parameter, the following restart type information:
(Bug#47366)
Bugs fixed:
Node takeover during a system restart occurs when the REDO log
for one or more data nodes is out of date, so that a node
restart is invoked for that node or those nodes. If this happens
while a mysqld process is attached to the
cluster as an SQL node, the mysqld takes a
global schema lock (a row lock), while trying to set up
cluster-internal replication.
However, this setup process could fail, causing the global
schema lock to be held for an excessive length of time, which
made the node restart hang as well. As a result, the mysqld
failed to set up cluster-internal replication, which led to
tables being read-only, and caused one node to hang during the
restart.
Note
This issue could actually occur in MySQL Cluster NDB 7.0 only,
but the fix was also applied MySQL Cluster NDB 6.3, in order
to keep the two codebases in alignment.
(Bug#49560)
Sending SIGHUP to a mysqld
running with the --ndbcluster and
--log-bin options caused the
process to crash instead of refreshing its log files.
(Bug#49515)
If the master data node receiving a request from a newly-started
API or data node for a node ID died before the request has been
handled, the management server waited (and kept a mutex) until
all handling of this node failure was complete before responding
to any other connections, instead of responding to other
connections as soon as it was informed of the node failure (that
is, it waited until it had received a NF_COMPLETEREP signal
rather than a NODE_FAILREP signal). On visible effect of this
misbehavior was that it caused management client commands such
as SHOW and ALL STATUS to respond with unnecessary slowness in
such circumstances.
(Bug#49207)
Attempting to create more than 11435 tables failed with Error
306 (Out of fragment records in DIH).
(Bug#49156)
When evaluating the options
--include-databases ,
--include-tables ,
--exclude-databases , and
--exclude-tables , the
ndb_restore program overwrote the result of
the database-level options with the result of the table-level
options rather than merging these results together, sometimes
leading to unexpected and unpredictable results.
As part of the fix for this problem, the semantics of these
options have been clarified; because of this, the rules
governing their evaluation have changed slightly. These changes
be summed up as follows:
All --include-* and
--exclude-* options are now evaluated from
right to left in the order in which they are passed to
ndb_restore.
All --include-* and
--exclude-* options are now cumulative.
In the event of a conflict, the first (rightmost) option
takes precedence.
For more detailed information and examples, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#48907)
When performing tasks that generated large amounts of I/O (such
as when using ndb_restore), an internal
memory buffer could overflow, causing data nodes to fail with
signal 6.
Subsequent analysis showed that this buffer was not actually
required, so this fix removes it.
(Bug#48861)
Exhaustion of send buffer memory or long signal memory caused
data nodes to crash. Now an appropriate error message is
provided instead when this situation occurs.
(Bug#48852)
In some situations, when it was not possible for an SQL node to
start a schema transaction (necessary, for instance, as part of
an online ALTER TABLE ),
NDBCLUSTER did not correctly
indicate the error to the MySQL server, which led
mysqld to crash.
(Bug#48841)
Under certain conditions, accounting of the number of free scan
records in the local query handler could be incorrect, so that
during node recovery or a local checkpoint operations, the LQH
could find itself lacking a scan record that is expected to
find, causing the node to crash.
(Bug#48697) See also Bug#48564.
The creation of an ordered index on a table undergoing DDL
operations could cause a data node crash under certain
timing-dependent conditions.
(Bug#48604)
During an LCP master takeover, when the newly elected master did
not receive a COPY_GCI LCP protocol message
but other nodes participating in the local checkpoint had
received one, the new master could use an uninitialized
variable, which caused it to crash.
(Bug#48584)
When running many parallel scans, a local checkpoint (which
performs a scan internally) could find itself not getting a scan
record, which led to a data node crash. Now an extra scan record
is reserved for this purpose, and a problem with obtaining the
scan record returns an appropriate error (error code 489,
Too many active scans).
(Bug#48564)
During a node restart, logging was enabled on a per-fragment
basis as the copying of each fragment was completed but local
checkpoints were not enabled until all fragments were copied,
making it possible to run out of redo log file space
(NDB error code 410) before the
restart was complete. Now logging is enabled only after all
fragments has been copied, just prior to enabling local
checkpoints.
(Bug#48474)
When using very large transactions containing many inserts,
ndbmtd could fail with Signal
11 without an easily detectable reason, due to an
internal variable being unitialized in the event that the
LongMessageBuffer was overloaded. Now, the
variable is initialized in such cases, avoiding the crash, and
an appropriate error message is generated.
(Bug#48441) See also Bug#46914.
A data node crashing while restarting, followed by a system
restart could lead to incorrect handling of redo log metadata,
causing the system restart to fail with Error while
reading REDO log.
(Bug#48436)
Starting a mysqld process with
--ndb-nodeid (either as
a command-line option or by assigning it a value in
my.cnf ) caused the
mysqld to get only the corresponding
connection from the [mysqld] section in the
config.ini file having the matching ID,
even when connection pooling was enabled (that is, when the
mysqld process was started with
--ndb-cluster-connection-pool set
greater than 1).
(Bug#48405) See also Bug#27644, Bug#38590, Bug#41592.
The configuration check that each management server runs to
verify that all connected ndb_mgmd processes
have the same configuration could fail when a configuration
change took place while this check was in progress. Now in such
cases, the configuration check is rescheduled for a later time,
after the change is complete.
(Bug#48143)
When employing NDB native backup to
back up and restore an empty NDB
table that used a non-sequential
AUTO_INCREMENT value, the
AUTO_INCREMENT value was not restored
correctly.
(Bug#48005)
ndb_config --xml
--configinfo now indicates that
parameters belonging in the [SCI] ,
[SCI DEFAULT] , [SHM] , and
[SHM DEFAULT] sections of the
config.ini file are deprecated or
experimental, as appropriate.
(Bug#47365)
NDB stores blob column data in a
separate, hidden table that is not accessible from MySQL. If
this table was missing for some reason (such as accidental
deletion of the file corresponding to the hidden table) when
making a MySQL Cluster native backup, ndb_restore crashed when
attempting to restore the backup. Now in such cases, ndb_restore
fails with the error message Table
table_name has blob column
(column_name ) with missing parts
table in backup instead.
(Bug#47289)
In MySQL Cluster NDB 7.0, ndb_config and
ndb_error_reporter were printing warnings
about management and data nodes running on the same host to
stdout instead of stderr ,
as was the case in earlier MySQL Cluster release series.
(Bug#44689, Bug#49160) See also Bug#25941.
DROP DATABASE failed when there
were stale temporary NDB tables in
the database. This situation could occur if
mysqld crashed during execution of a
DROP TABLE statement after the
table definition had been removed from
NDBCLUSTER but before the
corresponding .ndb file had been removed
from the crashed SQL node's data directory. Now, when
mysqld executes DROP
DATABASE , it checks for these files and removes them
if there are no corresponding table definitions for them found
in NDBCLUSTER .
(Bug#44529)
Creating an NDB table with an
excessive number of large BIT
columns caused the cluster to fail. Now, an attempt to create
such a table is rejected with error 791 (Too many
total bits in bitfields).
(Bug#42046) See also Bug#42047.
When a long-running transaction lasting long enough to cause
Error 410 (REDO log files overloaded) was
later committed or rolled back, it could happen that
NDBCLUSTER was not able to release
the space used for the REDO log, so that the error condition
persisted indefinitely.
The most likely cause of such transactions is a bug in the
application using MySQL Cluster. This fix should handle most
cases where this might occur.
(Bug#36500)
Deprecation and usage information obtained from
ndb_config --configinfo
regarding the PortNumber and
ServerPort configuration parameters was
improved.
(Bug#24584) Disk Data:
When running a write-intensive workload with a very large disk
page buffer cache, CPU usage approached 100% during a local
checkpoint of a cluster containing Disk Data tables.
(Bug#49532) Disk Data:
NDBCLUSTER failed to provide a
valid error message it failed to commit schema transactions
during an initial start if the cluster was configured using the
InitialLogFileGroup parameter.
(Bug#48517) Disk Data:
In certain limited cases, it was possible when the cluster
contained Disk Data tables for ndbmtd to
crash during a system restart.
(Bug#48498) See also Bug#47832. Disk Data:
Repeatedly creating and then dropping Disk Data tables could
eventually lead to data node failures.
(Bug#45794, Bug#48910) Disk Data:
When a crash occurs due to a problem in Disk Data code, the
currently active page list is printed to
stdout (that is, in one or more
ndb_nodeid _out.log
files). One of these lists could contain an endless loop; this
caused a printout that was effectively never-ending. Now in such
cases, a maximum of 512 entries is printed from each list.
(Bug#42431) Disk Data:
When the FileSystemPathUndoFiles
configuration parameter was set to an non-existent path, the
data nodes shut down with the generic error code 2341
(Internal program error). Now in such
cases, the error reported is error 2815 (File not
found).
Cluster API:
When a DML operation failed due to a uniqueness violation on an
NDB table having more than one
unique index, it was difficult to determine which constraint
caused the failure; it was necessary to obtain an
NdbError object, then decode its
details property, which in could lead to
memory management issues in application code.
To help solve this problem, a new API method
Ndb::getNdbErrorDetail() is added, providing
a well-formatted string containing more precise information
about the index that caused the unque constraint violation. The
following additional changes are also made in the NDB API:
For more information, see
Ndb::getNdbErrorDetail() ,
The NdbError Structure, and
Dictionary::listObjects() .
(Bug#48851) Cluster API:
When using blobs, calling getBlobHandle()
requires the full key to have been set using
equal() , because
getBlobHandle() must access the key for
adding blob table operations. However, if
getBlobHandle() was called without first
setting all parts of the primary key, the application using it
crashed. Now, an appropriate error code is returned instead.
(Bug#28116, Bug#48973)
Changes in MySQL Cluster NDB 7.0.9a (5.1.39-ndb-7.0.9a) Bugs fixed: Changes in MySQL Cluster NDB 7.0.8a (5.1.37-ndb-7.0.8a) Bugs fixed:
The disconnection of an API or SQL node having a node ID greater
than 49 caused a forced shutdown of the cluster.
(Bug#47844)
The error message text for NDB
error code 410 (REDO log files
overloaded...) was truncated.
(Bug#23662)
Changes in MySQL Cluster NDB 7.0.7 (5.1.35-ndb-7.0.7) Functionality added or changed: Important Change:
The default value of the DiskIOThreadPool
data node configuration parameter has changed from 8 to 2.
On Solaris platforms, the MySQL Cluster management server and
NDB API applications now use CLOCK_REALTIME
as the default clock.
(Bug#46183)
Formerly, node IDs were represented in the cluster log using a
complex hexadecimal/binary encoding scheme. Now, node IDs are
reported in the cluster log using numbers in standard decimal
notation.
(Bug#44248)
A new option --exclude-missing-columns has been
added for the ndb_restore program. In the
event that any tables in the database or databases being
restored to have fewer columns than the same-named tables in the
backup, the extra columns in the backup's version of the
tables are ignored. For more information, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#43139) Note
This issue, originally resolved in MySQL 5.1.16, re-occurred
due to a later (unrelated) change. The fix has been
re-applied.
(Bug#25984)
Previously, it was possible to disable arbitration only by
setting ArbitrationRank to 0 on all
management and API nodes. A new data node configuration
parameter Arbitration simplifies this task;
to disable arbitration, you can now use Arbitration =
Disabled in the [ndbd default]
section of the config.ini file.
It is now also possible to configure arbitration in such a way
that the cluster waits until the time determined by
ArbitrationTimeout passes for an external
manager to perform arbitration instead of handling it
internally. This can be done by setting Arbitration =
WaitExternal in the [ndbd default]
section of the config.ini file.
The default value for the Arbitration parameter is
Default , which allows arbitration to proceed
normally, as determined by the
ArbitrationRank settings for the management
and API nodes.
For more information, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
Bugs fixed: Packaging:
The pkg installer for MySQL Cluster on
Solaris did not perform a complete installation due to an
invalid directory reference in the post-install script.
(Bug#41998)
The output from ndb_config
--configinfo
--xml
contained quote characters (" ) within quoted
XML attributes, causing it to be not well-formed.
(Bug#46891)
When using multi-threaded data node processes
(ndbmtd), it was possible for LQH threads to
continue running even after all NDB
tables had been dropped. This meant that dropping the last
remaining NDB table during a local
checkpoint could cause multi-threaded data nodes to fail.
(Bug#46890)
During a global checkpoint, LQH threads could run unevenly,
causing a circular buffer oveflow by the Subscription Manager,
which led to data node failure.
(Bug#46782) See also Bug#46123, Bug#46723, Bug#45612.
Restarting the cluster following a local checkpoint and an
online ALTER TABLE on a non-empty
table caused data nodes to crash.
(Bug#46651)
A combination of index creation and drop operations (or creating
and dropping tables having indexes) with node and system
restarts could lead to a crash.
(Bug#46552)
Following an upgrade from MySQL Cluster NDB 6.3.x to MySQL
Cluster NDB 7.0.6, DDL and backup operations failed.
(Bug#46494, Bug#46563)
Full table scans failed to execute when the cluster contained
more than 21 table fragments.
Note
The number of table fragments in the cluster can be calculated
as the number of data nodes, times 8 (that is, times the value
of the internal constant
MAX_FRAG_PER_NODE ), divided by the number
of replicas. Thus, when NoOfReplicas = 1 at
least 3 data nodes were required to trigger this issue, and
when NoOfReplicas = 2 at least 4 data nodes
were required to do so.
(Bug#46490)
Killing MySQL Cluster nodes immediately following a local
checkpoint could lead to a crash of the cluster when later
attempting to perform a system restart.
The exact sequence of events causing this issue was as follows:
Local checkpoint occurs.
Immediately following the LCP, kill the master data node.
Kill the remaining data nodes within a few seconds of
killing the master.
Attempt to restart the cluster.
(Bug#46412)
Creating an index when the cluster had run out of table records
could cause data nodes to crash.
(Bug#46295)
Ending a line in the config.ini file with
an extra semicolon character (; ) caused
reading the file to fail with a parsing error.
(Bug#46242)
When combining an index scan and a delete with a primary key
delete, the index scan and delete failed to initialize a flag
properly. This could in rare circumstances cause a data node to
crash.
(Bug#46069)
OPTIMIZE TABLE on an
NDB table could in some cases cause
SQL and data nodes to crash. This issue was observed with both
ndbd and ndbmtd.
(Bug#45971)
The AutoReconnect configuration parameter for
API nodes (including SQL nodes) has been added. This is intended
to prevent API nodes from re-using allocated node IDs during
cluster restarts. For more information, see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
This fix also introduces two new methods of the
Ndb_cluster_connection class in the NDB API.
For more information, see
Ndb_cluster_connection::set_auto_reconnect() ,
and
Ndb_cluster_connection::get_auto_reconnect() .
(Bug#45921)
DML statements run during an upgrade from MySQL Cluster NDB 6.3
to NDB 7.0 were not handled correctly.
(Bug#45917)
On Windows, the internal
basestring_vsprintf() function did not return
a POSIX-compliant value as expected, causing the management
server to crash when trying to start a MySQL Cluster with more
than 4 data nodes.
(Bug#45733)
The signals used by ndb_restore to send
progress information about backups to the cluster log accessed
the cluster transporter without using any locks. Because of
this, it was theoretically possible that these signals could be
interefered with by heartbeat signals if both were sent at the
same time, causing the ndb_restore messages
to be corrupted.
(Bug#45646)
Due to changes in the way that
NDBCLUSTER handles schema changes
(implementation of schema transactions) in MySQL Cluster NDB
7.0, it was not possible to create MySQL Cluster tables having
more than 16 indexes using a single CREATE
TABLE statement.
This issue occurs only in MySQL Cluster NDB 7.0 releases prior
to 7.0.7 (including releases numbered NDB 6.4.x).
If you are not yet able to upgrade from an earlier MySQL Cluster
NDB 7.0 release, you can work around this problem by creating
the table without any indexes, then adding the indexes using a
separate CREATE INDEX statement
for each index.
(Bug#45525)
storage/ndb/src/common/util/CMakeLists.txt
did not build the BaseString-t test program
for Windows as the equivalent
storage/ndb/src/common/util/Makefile.am
does when building MySQL Cluster on Unix platforms.
(Bug#45099)
Problems could arise when using
VARCHAR columns
whose size was greater than 341 characters and which used the
utf8_unicode_ci collation. In some cases,
this combination of conditions could cause certain queries and
OPTIMIZE TABLE statements to
crash mysqld.
(Bug#45053)
The warning message Possible bug in
Dbdih::execBLOCK_COMMIT_ORD ... could sometimes
appear in the cluster log. This warning is obsolete, and has
been removed.
(Bug#44563)
Debugging code causing ndbd to use file
compression on NTFS filesystems failed with an error. (The code
was removed.) This issue affected debug builds of MySQL Cluster
on Windows platforms only.
(Bug#44418)
ALTER TABLE
REORGANIZE PARTITION could fail with Error 741
(Unsupported alter table) if the
appropriate hash-map was not present. This could occur when
adding nodes online; for example, when going from 2 data nodes
to 3 data nodes with NoOfReplicas=1 , or from
4 data nodes to 6 data nodes with
NoOfReplicas=2 .
(Bug#44301)
Previously, a GCP STOP event was written to
the cluster log as an INFO event. Now it is
logged as a WARNING event instead.
(Bug#43853)
In some cases, OPTIMIZE TABLE on
an NDB table did not free any
DataMemory .
(Bug#43683)
If the cluster crashed during the execution of a
CREATE LOGFILE GROUP statement,
the cluster could not be restarted afterwards.
(Bug#36702) See also Bug#34102. Disk Data: Partitioning:
An NDBCLUSTER table created with a
very large value for the MAX_ROWS option
could — if this table was dropped and a new table with
fewer partitions, but having the same table ID, was created
— cause ndbd to crash when performing a
system restart. This was because the server attempted to examine
each partition whether or not it actually existed.
(Bug#45154) Disk Data:
If the value set in the config.ini file for
FileSystemPathDD ,
FileSystemPathDataFiles , or
FileSystemPathUndoFiles was identical to the
value set for FileSystemPath , that parameter
was ignored when starting the data node with
--initial option. As a result, the Disk Data
files in the corresponding directory were not removed when
performing an initial start of the affected data node or data
nodes.
(Bug#46243)
Changes in MySQL Cluster NDB 7.0.6 (5.1.34-ndb-7.0.6) Functionality added or changed:
The ndb_config utility program can now
provide an offline dump of all MySQL Cluster configuration
parameters including information such as default and permitted
values, brief description, and applicable section of the
config.ini file. A dump in text format is
produced when running ndb_config with the new
--configinfo option, and in XML format when the
options --configinfo --xml are used together.
For more information and examples, see
Section 17.4.6, “ndb_config — Extract MySQL Cluster Configuration Information”.
Bugs fixed: Important Change: Partitioning:
User-defined partitioning of an
NDBCLUSTER table without any
primary key sometimes failed, and could cause
mysqld to crash.
Now, if you wish to create an
NDBCLUSTER table with user-defined
partitioning, the table must have an explicit primary key, and
all columns listed in the partitioning expression must be part
of the primary key. The hidden primary key used by the
NDBCLUSTER storage engine is not
sufficient for this purpose. However, if the list of columns is
empty (that is, the table is defined using PARTITION BY
[LINEAR] KEY() ), then no explicit primary key is
required.
This change does not effect the partitioning of tables using any
storage engine other than
NDBCLUSTER .
(Bug#40709) Important Change:
Previously, the configuration parameter
NoOfReplicas had no default value. Now the
default for NoOfReplicas is 2, which is the
recommended value in most settings.
(Bug#44746) Important Note:
It was not possible to perform an online upgrade from any MySQL
Cluster NDB 6.x release to MySQL Cluster NDB 7.0.5 or any to
earlier MySQL Cluster NDB 7.0 release.
With this fix, it is possible in MySQL Cluster NDB 7.0.6 and
later to perform online upgrades from MySQL Cluster NDB 6.3.8
and later MySQL Cluster NDB 6.3 releases, or from MySQL Cluster
NDB 7.0.5 or later MySQL Cluster NDB 7.0 releases. Online
upgrades to MySQL Cluster NDB 7.0 releases previous to MySQL
Cluster NDB 7.0.6 from earlier MySQL Cluster releases remain
unsupported; online upgrades from MySQL Cluster NDB 7.0 releases
previous to MySQL Cluster NDB 7.0.5 (including NDB 6.4.x beta
releases) to later MySQL Cluster NDB 7.0 releases also remain
unsupported.
(Bug#44294)
An internal NDB API buffer was not properly initialized.
(Bug#44977)
When a data node had written its GCI marker to the first page of
a megabyte, and that node was later killed during restart after
having processed that page (marker) but before completing a LCP,
the data node could fail with filesystem errors.
(Bug#44952) See also Bug#42564, Bug#44291.
When restarting a data nodes, management and API nodes
reconnecting to it failed to re-use existing ports that had
already been dynamically allocated for communications with that
data node.
(Bug#44866)
When ndb_config could not find the file
referenced by the --config-file option, it
tried to read my.cnf instead, then failed
with a misleading error message.
(Bug#44846)
When a data node was down so long that its most recent local
checkpoint depended on a global checkpoint that was no longer
restorable, it was possible for it to be unable to use optimized
node recovery when being restarted later.
(Bug#44844) See also Bug#26913.
Online upgrades to MySQL Cluster NDB 7.0 from a MySQL Cluster
NDB 6.3 release could fail due to changes in the handling of key
lengths and unique indexes during node recovery.
(Bug#44827)
ndb_config
--xml
did not output any entries for the HostName
parameter. In addition, the default listed for
MaxNoOfFiles was outside the allowed range of
values.
(Bug#44749) See also Bug#44685, Bug#44746.
The output of ndb_config
--xml
did not provide information about all sections of the
configuration file.
(Bug#44685) See also Bug#44746, Bug#44749.
Use of __builtin_expect() had the side effect
that compiler warnings about misuse of =
(assignment) instead of == in comparisons
were lost when building in debug mode. This is no longer
employed when configuring the build with the
--with-debug option.
(Bug#44570) See also Bug#44567.
Inspection of the code revealed that several assignment
operators (= ) were used in place of
comparison operators (== ) in
DbdihMain.cpp .
(Bug#44567) See also Bug#44570.
When using large numbers of configuration parameters, the
management server took an excessive amount of time (several
minutes or more) to load these from the configuration cache when
starting. This problem occurred when there were more than 32
configuration parameters specified, and became progressively
worse with each additional multiple of 32 configuration
parameters.
(Bug#44488)
Building the MySQL Cluster NDB 7.0 tree failed when using the
icc compiler.
(Bug#44310)
SSL connections to SQL nodes failed on big-endian platforms.
(Bug#44295)
Signals providing node state information
(NODE_STATE_REP and
CHANGE_NODE_STATE_REQ ) were not propagated to
all blocks of ndbmtd. This could cause the
following problems:
Inconsistent redo logs when performing a graceful shutdown;
Data nodes crashing when later restarting the cluster, data
nodes needing to perform node recovery during the system
restart, or both.
(Bug#44291) See also Bug#42564.
An NDB internal timing function did not work correctly on
Windows and could cause mysqld to fail on
some AMD processors, or when running inside a virtual machine.
(Bug#44276)
It was possible for NDB API applications to insert corrupt data
into the database, which could subquently lead to data node
crashes. Now, stricter checking is enforced on input data for
inserts and updates.
(Bug#44132)
ndb_restore failed when trying to restore
data on a big-endian machine from a backup file created on a
little-endian machine.
(Bug#44069)
Repeated starting and stopping of data nodes could cause
ndb_mgmd to fail. This issue was observed on Solaris/SPARC.
(Bug#43974)
A number of incorrectly formatted output strings in the source
code caused compiler warnings.
(Bug#43878)
When trying to use a data node with an older version of the
management server, the data node crashed on startup.
(Bug#43699)
In some cases, data node restarts during a system restart could
fail due to insufficient redo log space.
(Bug#43156)
NDBCLUSTER did not build correctly
on Solaris 9 platforms.
(Bug#39080) See also Bug#39036, Bug#39038.
The output of ndbd --help
did not provide clear information about the program's
--initial and --initial-start
options.
(Bug#28905)
It was theoretically possible for the value of a nonexistent
column to be read as NULL , rather than
causing an error.
(Bug#27843) Disk Data:
During a checkpoint, restore points are created for both the
on-disk and in-memory parts of a Disk Data table. Under certain
rare conditions, the in-memory restore point could include or
exclude a row that should have been in the snapshot. This would
later lead to a crash during or following recovery.
This issue was somewhat more likely to be encountered when using
ndbmtd.
(Bug#41915) See also Bug#47832. Disk Data:
This fix supercedes and improves on an earlier fix made for this
bug in MySQL 5.1.18.
(Bug#24521)
Changes in MySQL Cluster NDB 7.0.5 (5.1.32-ndb-7.0.5) Functionality added or changed:
Two new server status variables
Ndb_scan_count and
Ndb_pruned_scan_count have
been introduced.
Ndb_scan_count gives the
number of scans executed since the cluster was last started.
Ndb_pruned_scan_count gives
the number of scans for which
NDBCLUSTER was able to use
partition pruning. Together, these variables can be used to help
determine in the MySQL server whether table scans are pruned by
NDBCLUSTER .
(Bug#44153)
Bugs fixed: Important Note:
Due to problem discovered after the code freeze for this
release, it is not possible to perform an online upgrade from
any MySQL Cluster NDB 6.x release to MySQL Cluster NDB 7.0.5 or
any earlier MySQL Cluster NDB 7.0 release.
This issue is fixed in MySQL Cluster NDB 7.0.6 and later for
upgrades from MySQL Cluster NDB 6.3.8 and later MySQL Cluster
NDB 6.3 releases, or from MySQL Cluster NDB 7.0.5.
(Bug#44294) Cluster Replication:
If data node failed during an event creation operation, there
was a slight risk that a surviving data node could send an
invalid table reference back to NDB, causing the operation to
fail with a false Error 723 (No such
table). This could take place when a data node
failed as a mysqld process was setting up
MySQL Cluster Replication.
(Bug#43754) Cluster API:
The following issues occurred when performing an online
(rolling) upgrade of a cluster to a version of MySQL Cluster
that supports configuration caching from a version that does
not:
When using multiple management servers, after upgrading and
restarting one ndb_mgmd, any remaining
management servers using the previous version of
ndb_mgmd could not synchronize their
configuration data.
The MGM API ndb_mgm_get_configuration()
function failed to obtain configuration data.
(Bug#43641) Cluster API:
The following issues occurred when performing an online
(rolling) upgrade of a cluster to a version of MySQL Cluster
that supports configuration caching from a version that does
not:
When using multiple management servers, after upgrading and
restarting one ndb_mgmd, any remaining
management servers using the previous version of
ndb_mgmd could not synchronize their
configuration data.
The MGM API ndb_mgm_get_configuration()
function failed to obtain configuration data.
(Bug#43641)
If the number of fragments per table rises above a certain
threshold, the DBDIH kernel block's
on-disk table-definition grows large enough to occupy 2 pages.
However, in MySQL Cluster NDB 7.0 (including MySQL Cluster NDB
6.4 releases), only 1 page was actually written, causing table
definitions stored on disk to be incomplete.
This issue was not observed in MySQL Cluster release series
prior to MySQL Cluster NDB 7.0.
(Bug#44135)
TransactionDeadlockDetectionTimeout values
less than 100 were treated as 100. This could cause scans to
time out unexpectedly.
(Bug#44099)
The file ndberror.c contained a C++-style
comment, which caused builds to fail with some C compilers.
(Bug#44036)
A race condition could occur when a data node failed to restart
just before being included in the next global checkpoint. This
could cause other data nodes to fail.
(Bug#43888)
The setting for
ndb_use_transactions was
ignored. This issue was only known to occur in MySQL Cluster NDB
6.4.3 and MySQL Cluster NDB 7.0.4.
(Bug#43236)
When a data node process had been killed after allocating a node
ID, but before making contact with any other data node
processes, it was not possible to restart it due to a node ID
allocation failure.
This issue could effect either ndbd or
ndbmtd processes.
(Bug#43224) This regression was introduced by Bug#42973.
ndb_restore crashed when trying to restore a
backup made to a MySQL Cluster running on a platform having
different endianness from that on which the original backup was
taken.
(Bug#39540)
PID files for the data and management node daemons were not
removed following a normal shutdown.
(Bug#37225)
ndb_restore --print_data did
not handle DECIMAL columns
correctly.
(Bug#37171)
Invoking the management client START BACKUP
command from the system shell (for example, as ndb_mgm
-e "START BACKUP") did not work correctly, unless the
backup ID was included when the command was invoked.
Now, the backup ID is no longer required in such cases, and the
backup ID that is automatically generated is printed to stdout,
similar to how this is done when invoking START
BACKUP within the management client.
(Bug#31754)
When aborting an operation involving both an insert and a
delete, the insert and delete were aborted separately. This was
because the transaction coordinator did not know that the
operations affected on same row, and, in the case of a
committed-read (tuple or index) scan, the abort of the insert
was performed first, then the row was examined after the insert
was aborted but before the delete was aborted. In some cases,
this would leave the row in a inconsistent state. This could
occur when a local checkpoint was performed during a backup.
This issue did not affect primary ley operations or scans that
used locks (these are serialized).
After this fix, for ordered indexes, all operations that follow
the operation to be aborted are now also aborted.
Disk Data:
When using multi-threaded data nodes, DROP
TABLE statements on Disk Data tables could hang.
(Bug#43825) Disk Data:
This fix completes one that was made for this issue in MySQL
Cluster NDB-7.0.4, which did not rectify the problem in all
cases.
(Bug#43632) Cluster API:
If the largest offset of a
RecordSpecification
used for an
NdbRecord
object was for the NULL bits (and thus not a
column), this offset was not taken into account when calculating
the size used for the RecordSpecification .
This meant that the space for the NULL bits
could be overwritten by key or other information.
(Bug#43891) Cluster API:
BIT columns created using the
native NDB API format that were not created as nullable could
still sometimes be overwritten, or cause other columns to be
overwritten.
This issue did not effect tables having
BIT columns created using the
mysqld format (always used by MySQL Cluster SQL nodes).
(Bug#43802)
Changes in MySQL Cluster NDB 7.0.4 (5.1.32-ndb-7.0.4) Functionality added or changed: Important Change:
The default values for a number of MySQL Cluster configuration
parameters relating to memory usage and buffering have changed.
These parameters include RedoBuffer ,
LongMessageBuffer ,
BackupMemory ,
BackupDataBufferSize ,
BackupLogBufferSize ,
BackupWriteSize ,
BackupMaxWriteSize ,
SendBufferMemory (when applied to TCP
transporters), and ReceiveBufferMemory .
For more information, see
Section 17.3, “MySQL Cluster Configuration”.
When restoring from backup, ndb_restore now
reports the last global checkpoint reached when the backup was
taken.
(Bug#37384)
Bugs fixed: Cluster API:
Partition pruning did not work correctly for queries involving
multiple range scans.
As part of the fix for this issue, several improvements have
been made in the NDB API, including the addition of a new
NdbScanOperation::getPruned() method, a new
variant of NdbIndexScanOperation::setBound() ,
and a new Ndb::PartitionSpec data structure.
For more information about these changes, see
NdbScanOperation::getPruned() ,
NdbIndexScanOperation::setBound , and
The PartitionSpec Structure.
(Bug#37934) Cluster API:
Partition pruning did not work correctly for queries involving
multiple range scans.
As part of the fix for this issue, several improvements have
been made in the NDB API, including the addition of a new
NdbScanOperation::getPruned() method, a new
variant of NdbIndexScanOperation::setBound() ,
and a new Ndb::PartitionSpec data structure.
For more information about these changes, see
NdbScanOperation::getPruned() ,
NdbIndexScanOperation::setBound , and
The PartitionSpec Structure.
(Bug#37934)
TimeBetweenLocalCheckpoints was measured from
the end of one local checkpoint to the beginning of the next,
rather than from the beginning of one LCP to the beginning of
the next. This meant that the time spent performing the LCP was
not taken into account when determining the
TimeBetweenLocalCheckpoints interval, so that
LCPs were not started often enough, possibly causing data nodes
to run out of redo log space prematurely.
(Bug#43567)
The management server failed to start correctly in daemon mode.
(Bug#43559)
Following a DROP NODEGROUP command, the
output of SHOW in the
ndb_mgm cliently was not updated to reflect
the fact that the data nodes affected by this command were no
longer part of a node group.
(Bug#43413)
Using indexes containing variable-sized columns could lead to
internal errors when the indexes were being built.
(Bug#43226)
When using ndbmtd, multiple data node
failures caused the remaining data nodes to fail as well.
(Bug#43109)
It was not possible to add new data nodes to the cluster online
using multi-threaded data node processes
(ndbmtd).
(Bug#43108)
Some queries using combinations of logical and comparison
operators on an indexed column in the WHERE
clause could fail with the error Got error 4541
'IndexBound has no bound information' from
NDBCLUSTER.
(Bug#42857) Disk Data:
When using multi-threaded data nodes, dropping a Disk Data table
followed by a data node restart led to a crash.
(Bug#43632) Disk Data:
When using ndbmtd, repeated high-volume
inserts (on the order of 10000 rows inserted at a time) on a
Disk Data table would eventually lead to a data node crash.
(Bug#41398) Disk Data:
When a log file group had an undo log file whose size was too
small, restarting data nodes failed with Read
underflow errors.
As a result of this fix, the minimum allowed
INTIAL_SIZE for an undo log file is now
1M (1 megabyte).
(Bug#29574) Cluster API:
The default NdbRecord structures created by
NdbDictionary could have overlapping null
bits and data fields.
(Bug#43590) Cluster API:
When performing insert or write operations,
NdbRecord allows key columns to be specified
in both the key record and in the attribute record. Only one key
column value for each key column should be sent to the NDB
kernel, but this was not guaranteed. This is now ensured as
follows: For insert and write operations, key column values are
taken from the key record; for scan takeover update operations,
key column values are taken from the attribute record.
(Bug#42238) Cluster API:
Ordered index scans using NdbRecord formerly
expressed a BoundEQ range as separate lower
and upper bounds, resulting in 2 copies of the column values
being sent to the NDB kernel.
Now, when a range is specified by
NdbScanOperation::setBound() , the passed
pointers, key lengths, and inclusive bits are compared, and only
one copy of the equal key columns is sent to the kernel. This
makes such operations more efficient, as half the amount of
KeyInfo is now sent for a
BoundEQ range as before.
(Bug#38793)
Changes in MySQL Cluster NDB 6.4.3 (5.1.32-ndb-6.4.3) Functionality added or changed:
A new data node configuration parameter
MaxLCPStartDelay has been introduced to
facilitate parallel node recovery by causing a local checkpoint
to be delayed while recovering nodes are synchronizing data
dictionaries and other meta-information. For more information
about this parameter, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
(Bug#43053)
New options are introduced for ndb_restore
for determining which tables or databases should be restored:
--include-tables and
--include-databases can be used to restore
specific tables or databases.
--exclude-tables and
--exclude-databases can be used to exclude
the specified tables or databases from being restored.
For more information about these options, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#40429) Disk Data:
It is now possible to specify default locations for Disk Data
data files and undo log files, either together or separately,
using the data node configuration parameters
FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . For information
about these configuration parameters, see
Disk
Data filesystem parameters.
It is also now possible to specify a log file group, tablespace,
or both, that is created when the cluster is started, using the
InitialLogFileGroup and
InitialTablespace data node configuration
parameters. For information about these configuration
parameters, see
Disk
Data object creation parameters.
Bugs fixed: Performance:
Updates of the SYSTAB_0 system table to
obtain a unique identifier did not use transaction hints for
tables having no primary key. In such cases the NDB kernel used
a cache size of 1. This meant that each insert into a table not
having a primary key required an update of the corresponding
SYSTAB_0 entry, creating a potential
performance bottleneck.
With this fix, inserts on NDB tables without
primary keys can be under some conditions be performed up to
100% faster than previously.
(Bug#39268) Important Note:
It is not possible in this release to install the
InnoDB plugin if
InnoDB support has been compiled
into mysqld.
(Bug#42610) This regression was introduced by Bug#29263. Packaging:
Packages for MySQL Cluster were missing the
libndbclient.so and
libndbclient.a files.
(Bug#42278) Partitioning:
Executing ALTER TABLE ... REORGANIZE
PARTITION on an
NDBCLUSTER table having only one
partition caused mysqld to crash.
(Bug#41945) See also Bug#40389.
Backup IDs greater than 231 were not
handled correctly, causing negative values to be used in backup
directory names and printouts.
(Bug#43042)
When using ndbmtd, NDB kernel threads could
hang while trying to start the data nodes with
LockPagesInMainMemory set to 1.
(Bug#43021)
When using multiple management servers and starting several API
nodes (possibly including one or more SQL nodes) whose
connectstrings listed the management servers in different order,
it was possible for 2 API nodes to be assigned the same node ID.
When this happened it was possible for an API node not to get
fully connected, consequently producing a number of errors whose
cause was not easily recognizable.
(Bug#42973)
When using multi-threaded data nodes,
IndexMemory ,
MaxNoOfLocalOperations , and
MaxNoOfLocalScans were effectively multiplied
by the number of local query handlers in use by each
ndbmtd instance.
(Bug#42765) See also Bug#42215.
ndb_error_reporter worked correctly only with
GNU tar. (With other versions of
tar, it produced empty archives.)
(Bug#42753)
Triggers on NDBCLUSTER tables
caused such tables to become locked.
(Bug#42751) See also Bug#16229, Bug#18135.
When performing more than 32 index or tuple scans on a single
fragment, the scans could be left hanging. This caused
unnecessary timeouts, and in addition could possibly lead to a
hang of an LCP.
(Bug#42559)
A data node failure that occurred between calls to
NdbIndexScanOperation::readTuples(SF_OrderBy)
and NdbTransaction::Execute() was not
correctly handled; a subsequent call to
nextResult() caused a null pointer to be
deferenced, leading to a segfault in mysqld.
(Bug#42545)
If the cluster configuration cache file was larger than 32K, the
management server would not start.
(Bug#42543)
Issuing SHOW GLOBAL STATUS LIKE 'NDB%' before
mysqld had connected to the cluster caused a
segmentation fault.
(Bug#42458)
When using ndbmtd for all data nodes,
repeated failures of one data node during DML operations caused
other data nodes to fail.
(Bug#42450)
Data node failures that occurred before all data nodes had
connected to the cluster were not handled correctly, leading to
additional data node failures.
(Bug#42422)
When using multi-threaded data nodes, their
DataMemory and IndexMemory
usage as reported was multiplied by the number of local query
handlers (worker threads), making it appear that much more
memory was being used than was actually the case.
(Bug#42215) See also Bug#42765.
Given a MySQL Cluster containing no data (that is, whose data
nodes had all been started using --initial , and
into which no data had yet been imported) and having an empty
backup directory, executing START BACKUP with
a user-specified backup ID caused the data nodes to crash.
(Bug#41031)
In some cases, NDB did not check
correctly whether tables had changed before trying to use the
query cache. This could result in a crash of the debug MySQL
server.
(Bug#40464) Disk Data:
It was not possible to add an in-memory column online to a table
that used a table-level or column-level STORAGE
DISK option. The same issue prevented ALTER
ONLINE TABLE ... REORGANIZE PARTITION from working on
Disk Data tables.
(Bug#42549) Disk Data:
Repeated insert and delete operations on disk-based tables could
lead to failures in the NDB Tablespace Manager
(TSMAN kernel block).
(Bug#40344) Disk Data:
Creating a Disk Data tablespace with a very large extent size
caused the data nodes to fail. The issue was observed when using
extent sizes of 100 MB and larger.
(Bug#39096) Disk Data:
Trying to execute a CREATE LOGFILE
GROUP statement using a value greater than
150M for UNDO_BUFFER_SIZE
caused data nodes to crash.
As a result of this fix, the upper limit for
UNDO_BUFFER_SIZE is now
600M ; attempting to set a higher value now
fails gracefully with an error.
(Bug#34102) See also Bug#36702. Disk Data:
When attempting to create a tablespace that already existed, the
error message returned was Table or index with given
name already exists.
(Bug#32662) Disk Data:
Using a path or filename longer than 128 characters for Disk
Data undo log files and tablespace data files caused a number of
issues, including failures of CREATE
LOGFILE GROUP , ALTER LOGFILE
GROUP , CREATE
TABLESPACE , and ALTER
TABLESPACE statements, as well as crashes of
management nodes and data nodes.
With this fix, the maximum length for path and file names used
for Disk Data undo log files and tablespace data files is now
the same as the maximum for the operating system.
(Bug#31769, Bug#31770, Bug#31772) Disk Data:
Attempting to perform a system restart of the cluster where
there existed a logfile group without and undo log files caused
the data nodes to crash.
Note
While issuing a CREATE LOGFILE
GROUP statement without an ADD
UNDOFILE option fails with an error in the MySQL
server, this situation could arise if an SQL node failed
during the execution of a valid CREATE
LOGFILE GROUP statement; it is also possible to
create a logfile group without any undo log files using the
NDB API.
(Bug#17614) Cluster API:
Some error messages from ndb_mgmd contained
newline (\n ) characters. This could break the
MGM API protocol, which uses the newline as a line separator.
(Bug#43104) Cluster API:
When using an ordered index scan without putting all key columns
in the read mask, this invalid use of the NDB API went
undetected, which resulted in the use of uninitialized memory.
(Bug#42591)
Changes in MySQL Cluster NDB 6.4.2 (5.1.31-ndb-6.4.2) Bugs fixed:
Connections using IPv6 were not handled correctly by
mysqld.
(Bug#42413) See also Bug#42412, Bug#38247.
When a cluster backup failed with Error 1304 (Node
node_id1 : Backup request from
node_id2 failed to start), no clear
reason for the failure was provided.
As part of this fix, MySQL Cluster now retries backups in the
event of sequence errors.
(Bug#42354) See also Bug#22698.
Issuing SHOW ENGINE
NDBCLUSTER STATUS on an SQL node before the management
server had connected to the cluster caused
mysqld to crash.
(Bug#42264)
When using ndbmtd, setting
MaxNoOfThreads to a value higher than the
actual number of cores available and with insufficient
SharedGlobalMemory caused the data nodes to
crash.
The fix for this issue changes the behavior of
ndbmtd such that its internal job buffers no
longer rely on SharedGlobalMemory .
(Bug#42254)
Changes in MySQL Cluster NDB 6.4.1 (5.1.31-ndb-6.4.1) Functionality added or changed: Important Change:
Formerly, when the management server failed to create a
transporter for a data node connection,
net_write_timeout seconds
elapsed before the data node was actually allowed to disconnect.
Now in such cases the disconnection occurs immediately.
(Bug#41965) See also Bug#41713.
Formerly, when using MySQL Cluster Replication, records for
“empty” epochs — that is, epochs in which no
changes to NDBCLUSTER data or
tables took place — were inserted into the
ndb_apply_status and
ndb_binlog_index tables on the slave even
when --log-slave-updates was
disabled. Beginning with MySQL Cluster NDB 6.2.16 and MySQL
Cluster NDB 6.3.13 this was changed so that these
“empty” eopchs were no longer logged. However, it
is now possible to re-enable the older behavior (and cause
“empty” epochs to be logged) by using the
--ndb-log-empty-epochs option. For more
information, see Section 16.1.3.3, “Replication Slave Options and Variables”.
See also Bug#37472.
Bugs fixed:
A maximum of 11 TUP scans were allowed in
parallel.
(Bug#42084)
The management server could hang after attempting to halt it
with the STOP command in the management
client.
(Bug#42056) See also Bug#40922.
When using ndbmtd, one thread could flood
another thread, which would cause the system to stop with a
job buffer full condition (currently
implemented as an abort). This could be caused by committing or
aborting a large transaction (50000 rows or more) on a single
data node running ndbmtd. To prevent this
from happening, the number of signals that can be accepted by
the system threads is calculated before excuting them, and only
executing them if sufficient space is found.
(Bug#42052)
MySQL Cluster would not compile when using
libwrap . This issue was known to occur only
in MySQL Cluster NDB 6.4.0.
(Bug#41918)
Trying to execute an
ALTER ONLINE TABLE
... ADD COLUMN statement while inserting rows into the
table caused mysqld to crash.
(Bug#41905)
When a data node connects to the management server, the node
sends its node ID and transporter type; the management server
then verifies that there is a transporter set up for that node
and that it is in the correct state, and then sends back an
acknowledgement to the connecting node. If the transporter was
not in the correct state, no reply was sent back to the
connecting node, which would then hang until a read timeout
occurred (60 seconds). Now, if the transporter is not in the
correct state, the management server acknowledges this promptly,
and the node immediately disconnects.
(Bug#41713) See also Bug#41965.
Issuing EXIT in the management client
sometimes caused the client to hang.
(Bug#40922)
In the event that a MySQL Cluster backup failed due to file
permissions issues, conflicting reports were issued in the
management client.
(Bug#34526)
If all data nodes were shut down, MySQL clients were unable to
access NDBCLUSTER tables and data
even after the data nodes were restarted, unless the MySQL
clients themselves were restarted.
(Bug#33626)
Changes in MySQL Cluster NDB 6.4.0 (5.1.30-ndb-6.4.0) Functionality added or changed: Important Change:
MySQL Cluster now caches its configuration data. This means
that, by default, the management server only reads the global
configuration file (usually named
config.ini ) the first time that it is
started, and does not automatically re-read the this file when
restarted. This behavior can be controlled using new management
server options (--config-dir ,
--initial , and --reload ) that
have been added for this purpose. For more information, see
Section 17.3.2, “MySQL Cluster Configuration Files”, and
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
It is now possible while in Single User Mode to restart all data
nodes using ALL RESTART in the management
client. Restarting of individual nodes while in Single User Mode
remains disallowed.
(Bug#31056)
It is now possible to add data nodes to a MySQL Cluster online
— that is, to a running MySQL Cluster without shutting it
down.
For information about the procedure for adding data nodes
online, see Section 17.5.11, “Adding MySQL Cluster Data Nodes Online”.
A multi-threaded version of the MySQL Cluster data node daemon
is now available. The multi-threaded ndbmtd
binary is similar to ndbd and functions in
much the same way, but is intended for use on machines with
multiple CPU cores.
For more information, see
Section 17.4.3, “ndbmtd — The MySQL Cluster Data Node Daemon (Multi-Threaded)”.
It is now possible when performing a cluster backup to determine
whether the backup matches the state of the data when the backup
began or when it ended, using the new START
BACKUP options SNAPSHOTSTART and
SNAPSHOTEND in the management client. See
Section 17.5.3.2, “Using The MySQL Cluster Management Client to Create a Backup”,
for more information.
Bugs fixed:
API nodes disconnected too agressively from cluster when data
nodes were being restarted. This could sometimes lead to the API
node being unable to access the cluster at all during a rolling
restart.
(Bug#41462)
When long signal buffer exhaustion in the
ndbd process resulted in a signal being
dropped, the usual handling mechanism did not take fragmented
signals into account. This could result in a crash of the data
node because the fragmented signal handling mechanism was not
able to work with the missing fragments.
(Bug#39235)
The failure of a master node during a DDL operation caused the
cluster to be unavailable for further DDL operations until it
was restarted; failures of nonmaster nodes during DLL operations
caused the cluster to become completely inaccessible.
(Bug#36718)
Status messages shown in the management client when restarting a
management node were inappropriate and misleading. Now, when
restarting a management node, the messages displayed are as
follows, where node_id is the
management node's node ID:
ndb_mgm> node_id RESTART
Shutting down MGM node node_id for restart
Node node_id is being restarted
ndb_mgm>
(Bug#29275)
A data node failure when NoOfReplicas was
greater than 2 caused all cluster SQL nodes to crash.
(Bug#18621)
17.7.6.2. Changes in the MySQL Cluster NDB 6.3 Series
This section contains unified change history highlights for all
MySQL Cluster releases based on version 6.3 of the
NDBCLUSTER storage engine through
MySQL Cluster NDB 6.3.32. Included are all
changelog entries in the categories MySQL
Cluster, Disk Data, and
Cluster API.
For an overview of features that were added in MySQL Cluster NDB
6.3, see
Section 17.1.4.4, “MySQL Cluster Development in MySQL Cluster NDB 6.3”.
Changes in MySQL Cluster NDB 6.3.30 (5.1.39-ndb-6.3.30) Functionality added or changed: Changes in MySQL Cluster NDB 6.3.29 (5.1.39-ndb-6.3.29) Functionality added or changed:
This enhanced functionality is supported for upgrades to MySQL
Cluster NDB 7.0 when the NDB engine
version is 7.0.10 or later.
(Bug#48528, Bug#49163)
The output from ndb_config
--configinfo
--xml now indicates, for each
configuration parameter, the following restart type information:
(Bug#47366)
Bugs fixed:
Node takeover during a system restart occurs when the REDO log
for one or more data nodes is out of date, so that a node
restart is invoked for that node or those nodes. If this happens
while a mysqld process is attached to the
cluster as an SQL node, the mysqld takes a
global schema lock (a row lock), while trying to set up
cluster-internal replication.
However, this setup process could fail, causing the global
schema lock to be held for an excessive length of time, which
made the node restart hang as well. As a result, the mysqld
failed to set up cluster-internal replication, which led to
tables being read-only, and caused one node to hang during the
restart.
Note
This issue could actually occur in MySQL Cluster NDB 7.0 only,
but the fix was also applied MySQL Cluster NDB 6.3, in order
to keep the two codebases in alignment.
(Bug#49560)
Sending SIGHUP to a mysqld
running with the --ndbcluster and
--log-bin options caused the
process to crash instead of refreshing its log files.
(Bug#49515)
If the master data node receiving a request from a newly-started
API or data node for a node ID died before the request has been
handled, the management server waited (and kept a mutex) until
all handling of this node failure was complete before responding
to any other connections, instead of responding to other
connections as soon as it was informed of the node failure (that
is, it waited until it had received a NF_COMPLETEREP signal
rather than a NODE_FAILREP signal). On visible effect of this
misbehavior was that it caused management client commands such
as SHOW and ALL STATUS to respond with unnecessary slowness in
such circumstances.
(Bug#49207)
When evaluating the options
--include-databases ,
--include-tables ,
--exclude-databases , and
--exclude-tables , the
ndb_restore program overwrote the result of
the database-level options with the result of the table-level
options rather than merging these results together, sometimes
leading to unexpected and unpredictable results.
As part of the fix for this problem, the semantics of these
options have been clarified; because of this, the rules
governing their evaluation have changed slightly. These changes
be summed up as follows:
All --include-* and
--exclude-* options are now evaluated from
right to left in the order in which they are passed to
ndb_restore.
All --include-* and
--exclude-* options are now cumulative.
In the event of a conflict, the first (rightmost) option
takes precedence.
For more detailed information and examples, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#48907)
When performing tasks that generated large amounts of I/O (such
as when using ndb_restore), an internal
memory buffer could overflow, causing data nodes to fail with
signal 6.
Subsequent analysis showed that this buffer was not actually
required, so this fix removes it.
(Bug#48861)
Exhaustion of send buffer memory or long signal memory caused
data nodes to crash. Now an appropriate error message is
provided instead when this situation occurs.
(Bug#48852)
Under certain conditions, accounting of the number of free scan
records in the local query handler could be incorrect, so that
during node recovery or a local checkpoint operations, the LQH
could find itself lacking a scan record that is expected to
find, causing the node to crash.
(Bug#48697) See also Bug#48564.
The creation of an ordered index on a table undergoing DDL
operations could cause a data node crash under certain
timing-dependent conditions.
(Bug#48604)
During an LCP master takeover, when the newly elected master did
not receive a COPY_GCI LCP protocol message
but other nodes participating in the local checkpoint had
received one, the new master could use an uninitialized
variable, which caused it to crash.
(Bug#48584)
When running many parallel scans, a local checkpoint (which
performs a scan internally) could find itself not getting a scan
record, which led to a data node crash. Now an extra scan record
is reserved for this purpose, and a problem with obtaining the
scan record returns an appropriate error (error code 489,
Too many active scans).
(Bug#48564)
During a node restart, logging was enabled on a per-fragment
basis as the copying of each fragment was completed but local
checkpoints were not enabled until all fragments were copied,
making it possible to run out of redo log file space
(NDB error code 410) before the
restart was complete. Now logging is enabled only after all
fragments has been copied, just prior to enabling local
checkpoints.
(Bug#48474)
When employing NDB native backup to
back up and restore an empty NDB
table that used a non-sequential
AUTO_INCREMENT value, the
AUTO_INCREMENT value was not restored
correctly.
(Bug#48005)
ndb_config --xml
--configinfo now indicates that
parameters belonging in the [SCI] ,
[SCI DEFAULT] , [SHM] , and
[SHM DEFAULT] sections of the
config.ini file are deprecated or
experimental, as appropriate.
(Bug#47365)
NDB stores blob column data in a
separate, hidden table that is not accessible from MySQL. If
this table was missing for some reason (such as accidental
deletion of the file corresponding to the hidden table) when
making a MySQL Cluster native backup, ndb_restore crashed when
attempting to restore the backup. Now in such cases, ndb_restore
fails with the error message Table
table_name has blob column
(column_name ) with missing parts
table in backup instead.
(Bug#47289)
DROP DATABASE failed when there
were stale temporary NDB tables in
the database. This situation could occur if
mysqld crashed during execution of a
DROP TABLE statement after the
table definition had been removed from
NDBCLUSTER but before the
corresponding .ndb file had been removed
from the crashed SQL node's data directory. Now, when
mysqld executes DROP
DATABASE , it checks for these files and removes them
if there are no corresponding table definitions for them found
in NDBCLUSTER .
(Bug#44529)
Creating an NDB table with an
excessive number of large BIT
columns caused the cluster to fail. Now, an attempt to create
such a table is rejected with error 791 (Too many
total bits in bitfields).
(Bug#42046) See also Bug#42047.
When a long-running transaction lasting long enough to cause
Error 410 (REDO log files overloaded) was
later committed or rolled back, it could happen that
NDBCLUSTER was not able to release
the space used for the REDO log, so that the error condition
persisted indefinitely.
The most likely cause of such transactions is a bug in the
application using MySQL Cluster. This fix should handle most
cases where this might occur.
(Bug#36500)
Deprecation and usage information obtained from
ndb_config --configinfo
regarding the PortNumber and
ServerPort configuration parameters was
improved.
(Bug#24584) Disk Data:
When running a write-intensive workload with a very large disk
page buffer cache, CPU usage approached 100% during a local
checkpoint of a cluster containing Disk Data tables.
(Bug#49532) Disk Data:
Repeatedly creating and then dropping Disk Data tables could
eventually lead to data node failures.
(Bug#45794, Bug#48910) Disk Data:
When the FileSystemPathUndoFiles
configuration parameter was set to an non-existent path, the
data nodes shut down with the generic error code 2341
(Internal program error). Now in such
cases, the error reported is error 2815 (File not
found).
Cluster API:
When a DML operation failed due to a uniqueness violation on an
NDB table having more than one
unique index, it was difficult to determine which constraint
caused the failure; it was necessary to obtain an
NdbError object, then decode its
details property, which in could lead to
memory management issues in application code.
To help solve this problem, a new API method
Ndb::getNdbErrorDetail() is added, providing
a well-formatted string containing more precise information
about the index that caused the unque constraint violation. The
following additional changes are also made in the NDB API:
For more information, see
Ndb::getNdbErrorDetail() ,
The NdbError Structure, and
Dictionary::listObjects() .
(Bug#48851) Cluster API:
When using blobs, calling getBlobHandle()
requires the full key to have been set using
equal() , because
getBlobHandle() must access the key for
adding blob table operations. However, if
getBlobHandle() was called without first
setting all parts of the primary key, the application using it
crashed. Now, an appropriate error code is returned instead.
(Bug#28116, Bug#48973)
Changes in MySQL Cluster NDB 6.3.28a (5.1.39-ndb-6.3.28a) Bugs fixed: Changes in MySQL Cluster NDB 6.3.27 (5.1.37-ndb-6.3.27) Functionality added or changed: Bugs fixed:
mysqld allocated an excessively large buffer
for handling BLOB values due to
overestimating their size. (For each row, enough space was
allocated to accommodate every
BLOB or
TEXT column value in the result
set.) This could adversely affect performance when using tables
containing BLOB or
TEXT columns; in a few extreme
cases, this issue could also cause the host system to run out of
memory unexpectedly.
(Bug#47574) See also Bug#47572, Bug#47573.
NDBCLUSTER uses a dynamically-allocated
buffer to store BLOB or
TEXT column data that is read
from rows in MySQL Cluster tables.
When an instance of the NDBCLUSTER table
handler was recycled (this can happen due to table definition
cache pressure or to operations such as
FLUSH TABLES or
ALTER TABLE ), if the last row
read contained blobs of zero length, the buffer was not freed,
even though the reference to it was lost. This resulted in a
memory leak.
For example, consider the table defined and populated as shown
here:
CREATE TABLE t (a INT PRIMARY KEY, b LONGTEXT) ENGINE=NDB;
INSERT INTO t VALUES (1, REPEAT('F', 20000));
INSERT INTO t VALUES (2, '');
Now execute repeatedly a SELECT
on this table, such that the zero-length
LONGTEXT row is
last, followed by a FLUSH
TABLES statement (which forces the handler object to
be re-used), as shown here:
SELECT a, length(b) FROM bl ORDER BY a;
FLUSH TABLES;
Prior to the fix, this resulted in a memory leak proportional to
the size of the stored
LONGTEXT value
each time these two statements were executed.
(Bug#47573) See also Bug#47572, Bug#47574.
Large transactions involving joins between tables containing
BLOB columns used excessive
memory.
(Bug#47572) See also Bug#47573, Bug#47574.
A variable was left uninitialized while a data node copied data
from its peers as part of its startup routine; if the starting
node died during this phase, this could lead a crash of the
cluster when the node was later restarted.
(Bug#47505)
When a data node restarts, it first runs the redo log until
reaching the latest restorable global checkpoint; after this it
scans the remainder of the redo log file, searching for entries
that should be invalidated so they are not used in any
subsequent restarts. (It is possible, for example, if restoring
GCI number 25, that there might be entries belonging to GCI 26
in the redo log.) However, under certain rare conditions, during
the invalidation process, the redo log files themselves were not
always closed while scanning ahead in the redo log. In rare
cases, this could lead to MaxNoOfOpenFiles
being exceeded, causing a the data node to crash.
(Bug#47171)
For very large values of MaxNoOfTables +
MaxNoOfAttributes , the calculation for
StringMemory could overflow when creating
large numbers of tables, leading to NDB error 773
(Out of string memory, please modify StringMemory
config parameter), even when
StringMemory was set to
100 (100 percent).
(Bug#47170)
The default value for the StringMemory
configuration parameter, unlike other MySQL Cluster
configuration parameters, was not set in
ndb/src/mgmsrv/ConfigInfo.cpp .
(Bug#47166)
Signals from a failed API node could be received after an
API_FAILREQ signal (see
Operations and Signals)
has been received from that node, which could result in invalid
states for processing subsequent signals. Now, all pending
signals from a failing API node are processed before any
API_FAILREQ signal is received.
(Bug#47039) See also Bug#44607.
Using triggers on NDB tables caused
ndb_autoincrement_prefetch_sz
to be treated as having the NDB kernel's internal default
value (32) and the value for this variable as set on the
cluster's SQL nodes to be ignored.
(Bug#46712)
Running an ALTER TABLE statement
while an NDB backup was in progress caused
mysqld to crash.
(Bug#44695)
When performing auto-discovery of tables on individual SQL
nodes, NDBCLUSTER attempted to overwrite
existing MyISAM .frm
files and corrupted them.
Workaround.
In the mysql client, create a new table
(t2 ) with same definition as the corrupted
table (t1 ). Use your system shell or file
manager to rename the old .MYD file to
the new file name (for example, mv t1.MYD
t2.MYD). In the mysql client,
repair the new table, drop the old one, and rename the new
table using the old file name (for example,
RENAME TABLE t2
TO t1 ).
(Bug#42614)
Running ndb_restore with the
--print or --print_log option
could cause it to crash.
(Bug#40428, Bug#33040)
An insert on an NDB table was not
always flushed properly before performing a scan. One way in
which this issue could manifest was that
LAST_INSERT_ID() sometimes failed
to return correct values when using a trigger on an
NDB table.
(Bug#38034)
When a data node received a TAKE_OVERTCCONF
signal from the master before that node had received a
NODE_FAILREP , a race condition could in
theory result.
(Bug#37688) See also Bug#25364, Bug#28717.
Some joins on large NDB tables
having TEXT or
BLOB columns could cause
mysqld processes to leak memory. The joins
did not need to reference the
TEXT or
BLOB columns directly for this
issue to occur.
(Bug#36701)
On Mac OS X 10.5, commands entered in the management client
failed and sometimes caused the client to hang, although
management client commands invoked using the
--execute (or
-e ) option from the system shell worked
normally.
For example, the following command failed with an error and hung
until killed manually, as shown here:
ndb_mgm> SHOW
Warning, event thread startup failed, degraded printouts as result, errno=36
^C
However, the same management client command, invoked from the
system shell as shown here, worked correctly:
shell> ndb_mgm -e "SHOW"
(Bug#35751) See also Bug#34438. Disk Data:
Calculation of free space for Disk Data table fragments was
sometimes done incorrectly. This could lead to unnecessary
allocation of new extents even when sufficient space was
available in existing ones for inserted data. In some cases,
this might also lead to crashes when restarting data nodes.
Note
This miscalculation was not reflected in the contents of the
INFORMATION_SCHEMA.FILES table,
as it applied to extents allocated to a fragment, and not to a
file.
(Bug#47072) Cluster API:
In some circumstances, if an API node encountered a data node
failure between the creation of a transaction and the start of a
scan using that transaction, then any subsequent calls to
startTransaction() and
closeTransaction() could cause the same
transaction to be started and closed repeatedly.
(Bug#47329) Cluster API:
Performing multiple operations using the same primary key within
the same
NdbTransaction::execute()
call could lead to a data node crash.
Note
This fix does not make change the fact that performing
multiple operations using the same primary key within the same
execute() is not supported; because there
is no way to determine the order of such operations, the
result of such combined operations remains undefined.
(Bug#44065) See also Bug#44015.
Changes in MySQL Cluster NDB 6.3.26 (5.1.35-ndb-6.3.26) Functionality added or changed:
On Solaris platforms, the MySQL Cluster management server and
NDB API applications now use CLOCK_REALTIME
as the default clock.
(Bug#46183)
A new option --exclude-missing-columns has been
added for the ndb_restore program. In the
event that any tables in the database or databases being
restored to have fewer columns than the same-named tables in the
backup, the extra columns in the backup's version of the
tables are ignored. For more information, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#43139) Note
This issue, originally resolved in MySQL 5.1.16, re-occurred
due to a later (unrelated) change. The fix has been
re-applied.
(Bug#25984)
Bugs fixed:
Restarting the cluster following a local checkpoint and an
online ALTER TABLE on a non-empty
table caused data nodes to crash.
(Bug#46651)
Full table scans failed to execute when the cluster contained
more than 21 table fragments.
Note
The number of table fragments in the cluster can be calculated
as the number of data nodes, times 8 (that is, times the value
of the internal constant
MAX_FRAG_PER_NODE ), divided by the number
of replicas. Thus, when NoOfReplicas = 1 at
least 3 data nodes were required to trigger this issue, and
when NoOfReplicas = 2 at least 4 data nodes
were required to do so.
(Bug#46490)
Killing MySQL Cluster nodes immediately following a local
checkpoint could lead to a crash of the cluster when later
attempting to perform a system restart.
The exact sequence of events causing this issue was as follows:
Local checkpoint occurs.
Immediately following the LCP, kill the master data node.
Kill the remaining data nodes within a few seconds of
killing the master.
Attempt to restart the cluster.
(Bug#46412)
Ending a line in the config.ini file with
an extra semicolon character (; ) caused
reading the file to fail with a parsing error.
(Bug#46242)
When combining an index scan and a delete with a primary key
delete, the index scan and delete failed to initialize a flag
properly. This could in rare circumstances cause a data node to
crash.
(Bug#46069)
OPTIMIZE TABLE on an
NDB table could in some cases cause
SQL and data nodes to crash. This issue was observed with both
ndbd and ndbmtd.
(Bug#45971)
The AutoReconnect configuration parameter for
API nodes (including SQL nodes) has been added. This is intended
to prevent API nodes from re-using allocated node IDs during
cluster restarts. For more information, see
Section 17.3.2.7, “Defining SQL and Other API Nodes in a MySQL Cluster”.
This fix also introduces two new methods of the
Ndb_cluster_connection class in the NDB API.
For more information, see
Ndb_cluster_connection::set_auto_reconnect() ,
and
Ndb_cluster_connection::get_auto_reconnect() .
(Bug#45921)
The signals used by ndb_restore to send
progress information about backups to the cluster log accessed
the cluster transporter without using any locks. Because of
this, it was theoretically possible that these signals could be
interefered with by heartbeat signals if both were sent at the
same time, causing the ndb_restore messages
to be corrupted.
(Bug#45646)
Problems could arise when using
VARCHAR columns
whose size was greater than 341 characters and which used the
utf8_unicode_ci collation. In some cases,
this combination of conditions could cause certain queries and
OPTIMIZE TABLE statements to
crash mysqld.
(Bug#45053)
An internal NDB API buffer was not properly initialized.
(Bug#44977)
When a data node had written its GCI marker to the first page of
a megabyte, and that node was later killed during restart after
having processed that page (marker) but before completing a LCP,
the data node could fail with filesystem errors.
(Bug#44952) See also Bug#42564, Bug#44291.
The warning message Possible bug in
Dbdih::execBLOCK_COMMIT_ORD ... could sometimes
appear in the cluster log. This warning is obsolete, and has
been removed.
(Bug#44563)
In some cases, OPTIMIZE TABLE on
an NDB table did not free any
DataMemory .
(Bug#43683)
If the cluster crashed during the execution of a
CREATE LOGFILE GROUP statement,
the cluster could not be restarted afterwards.
(Bug#36702) See also Bug#34102. Disk Data: Partitioning:
An NDBCLUSTER table created with a
very large value for the MAX_ROWS option
could — if this table was dropped and a new table with
fewer partitions, but having the same table ID, was created
— cause ndbd to crash when performing a
system restart. This was because the server attempted to examine
each partition whether or not it actually existed.
(Bug#45154) Disk Data:
If the value set in the config.ini file for
FileSystemPathDD ,
FileSystemPathDataFiles , or
FileSystemPathUndoFiles was identical to the
value set for FileSystemPath , that parameter
was ignored when starting the data node with
--initial option. As a result, the Disk Data
files in the corresponding directory were not removed when
performing an initial start of the affected data node or data
nodes.
(Bug#46243) Disk Data:
During a checkpoint, restore points are created for both the
on-disk and in-memory parts of a Disk Data table. Under certain
rare conditions, the in-memory restore point could include or
exclude a row that should have been in the snapshot. This would
later lead to a crash during or following recovery.
(Bug#41915) See also Bug#47832.
Changes in MySQL Cluster NDB 6.3.25 (5.1.34-ndb-6.3.25) Functionality added or changed:
Two new server status variables
Ndb_scan_count and
Ndb_pruned_scan_count have
been introduced.
Ndb_scan_count gives the
number of scans executed since the cluster was last started.
Ndb_pruned_scan_count gives
the number of scans for which
NDBCLUSTER was able to use
partition pruning. Together, these variables can be used to help
determine in the MySQL server whether table scans are pruned by
NDBCLUSTER .
(Bug#44153)
The ndb_config utility program can now
provide an offline dump of all MySQL Cluster configuration
parameters including information such as default and permitted
values, brief description, and applicable section of the
config.ini file. A dump in text format is
produced when running ndb_config with the new
--configinfo option, and in XML format when the
options --configinfo --xml are used together.
For more information and examples, see
Section 17.4.6, “ndb_config — Extract MySQL Cluster Configuration Information”.
Bugs fixed: Important Change: Partitioning:
User-defined partitioning of an
NDBCLUSTER table without any
primary key sometimes failed, and could cause
mysqld to crash.
Now, if you wish to create an
NDBCLUSTER table with user-defined
partitioning, the table must have an explicit primary key, and
all columns listed in the partitioning expression must be part
of the primary key. The hidden primary key used by the
NDBCLUSTER storage engine is not
sufficient for this purpose. However, if the list of columns is
empty (that is, the table is defined using PARTITION BY
[LINEAR] KEY() ), then no explicit primary key is
required.
This change does not effect the partitioning of tables using any
storage engine other than
NDBCLUSTER .
(Bug#40709) Important Change:
Previously, the configuration parameter
NoOfReplicas had no default value. Now the
default for NoOfReplicas is 2, which is the
recommended value in most settings.
(Bug#44746) Packaging:
The pkg installer for MySQL Cluster on
Solaris did not perform a complete installation due to an
invalid directory reference in the post-install script.
(Bug#41998)
When ndb_config could not find the file
referenced by the --config-file option, it
tried to read my.cnf instead, then failed
with a misleading error message.
(Bug#44846)
When a data node was down so long that its most recent local
checkpoint depended on a global checkpoint that was no longer
restorable, it was possible for it to be unable to use optimized
node recovery when being restarted later.
(Bug#44844) See also Bug#26913.
ndb_config
--xml
did not output any entries for the HostName
parameter. In addition, the default listed for
MaxNoOfFiles was outside the allowed range of
values.
(Bug#44749) See also Bug#44685, Bug#44746.
The output of ndb_config
--xml
did not provide information about all sections of the
configuration file.
(Bug#44685) See also Bug#44746, Bug#44749.
Inspection of the code revealed that several assignment
operators (= ) were used in place of
comparison operators (== ) in
DbdihMain.cpp .
(Bug#44567) See also Bug#44570.
It was possible for NDB API applications to insert corrupt data
into the database, which could subquently lead to data node
crashes. Now, stricter checking is enforced on input data for
inserts and updates.
(Bug#44132)
ndb_restore failed when trying to restore
data on a big-endian machine from a backup file created on a
little-endian machine.
(Bug#44069)
The file ndberror.c contained a C++-style
comment, which caused builds to fail with some C compilers.
(Bug#44036)
When trying to use a data node with an older version of the
management server, the data node crashed on startup.
(Bug#43699)
In some cases, data node restarts during a system restart could
fail due to insufficient redo log space.
(Bug#43156)
NDBCLUSTER did not build correctly
on Solaris 9 platforms.
(Bug#39080) See also Bug#39036, Bug#39038.
ndb_restore --print_data did
not handle DECIMAL columns
correctly.
(Bug#37171)
The output of ndbd --help
did not provide clear information about the program's
--initial and --initial-start
options.
(Bug#28905)
It was theoretically possible for the value of a nonexistent
column to be read as NULL , rather than
causing an error.
(Bug#27843) Disk Data:
This fix supercedes and improves on an earlier fix made for this
bug in MySQL 5.1.18.
(Bug#24521)
Changes in MySQL Cluster NDB 6.3.24 (5.1.32-ndb-6.3.24) Bugs fixed: Cluster Replication:
If data node failed during an event creation operation, there
was a slight risk that a surviving data node could send an
invalid table reference back to NDB, causing the operation to
fail with a false Error 723 (No such
table). This could take place when a data node
failed as a mysqld process was setting up
MySQL Cluster Replication.
(Bug#43754) Cluster API:
Partition pruning did not work correctly for queries involving
multiple range scans.
As part of the fix for this issue, several improvements have
been made in the NDB API, including the addition of a new
NdbScanOperation::getPruned() method, a new
variant of NdbIndexScanOperation::setBound() ,
and a new Ndb::PartitionSpec data structure.
For more information about these changes, see
NdbScanOperation::getPruned() ,
NdbIndexScanOperation::setBound , and
The PartitionSpec Structure.
(Bug#37934) Cluster API:
Partition pruning did not work correctly for queries involving
multiple range scans.
As part of the fix for this issue, several improvements have
been made in the NDB API, including the addition of a new
NdbScanOperation::getPruned() method, a new
variant of NdbIndexScanOperation::setBound() ,
and a new Ndb::PartitionSpec data structure.
For more information about these changes, see
NdbScanOperation::getPruned() ,
NdbIndexScanOperation::setBound , and
The PartitionSpec Structure.
(Bug#37934)
TransactionDeadlockDetectionTimeout values
less than 100 were treated as 100. This could cause scans to
time out unexpectedly.
(Bug#44099)
A race condition could occur when a data node failed to restart
just before being included in the next global checkpoint. This
could cause other data nodes to fail.
(Bug#43888)
TimeBetweenLocalCheckpoints was measured from
the end of one local checkpoint to the beginning of the next,
rather than from the beginning of one LCP to the beginning of
the next. This meant that the time spent performing the LCP was
not taken into account when determining the
TimeBetweenLocalCheckpoints interval, so that
LCPs were not started often enough, possibly causing data nodes
to run out of redo log space prematurely.
(Bug#43567)
Using indexes containing variable-sized columns could lead to
internal errors when the indexes were being built.
(Bug#43226)
When a data node process had been killed after allocating a node
ID, but before making contact with any other data node
processes, it was not possible to restart it due to a node ID
allocation failure.
This issue could effect either ndbd or
ndbmtd processes.
(Bug#43224) This regression was introduced by Bug#42973.
Some queries using combinations of logical and comparison
operators on an indexed column in the WHERE
clause could fail with the error Got error 4541
'IndexBound has no bound information' from
NDBCLUSTER.
(Bug#42857)
ndb_restore crashed when trying to restore a
backup made to a MySQL Cluster running on a platform having
different endianness from that on which the original backup was
taken.
(Bug#39540)
When aborting an operation involving both an insert and a
delete, the insert and delete were aborted separately. This was
because the transaction coordinator did not know that the
operations affected on same row, and, in the case of a
committed-read (tuple or index) scan, the abort of the insert
was performed first, then the row was examined after the insert
was aborted but before the delete was aborted. In some cases,
this would leave the row in a inconsistent state. This could
occur when a local checkpoint was performed during a backup.
This issue did not affect primary ley operations or scans that
used locks (these are serialized).
After this fix, for ordered indexes, all operations that follow
the operation to be aborted are now also aborted.
Disk Data:
When a log file group had an undo log file whose size was too
small, restarting data nodes failed with Read
underflow errors.
As a result of this fix, the minimum allowed
INTIAL_SIZE for an undo log file is now
1M (1 megabyte).
(Bug#29574) Cluster API:
If the largest offset of a
RecordSpecification
used for an
NdbRecord
object was for the NULL bits (and thus not a
column), this offset was not taken into account when calculating
the size used for the RecordSpecification .
This meant that the space for the NULL bits
could be overwritten by key or other information.
(Bug#43891) Cluster API:
BIT columns created using the
native NDB API format that were not created as nullable could
still sometimes be overwritten, or cause other columns to be
overwritten.
This issue did not effect tables having
BIT columns created using the
mysqld format (always used by MySQL Cluster SQL nodes).
(Bug#43802) Cluster API:
The default NdbRecord structures created by
NdbDictionary could have overlapping null
bits and data fields.
(Bug#43590) Cluster API:
When performing insert or write operations,
NdbRecord allows key columns to be specified
in both the key record and in the attribute record. Only one key
column value for each key column should be sent to the NDB
kernel, but this was not guaranteed. This is now ensured as
follows: For insert and write operations, key column values are
taken from the key record; for scan takeover update operations,
key column values are taken from the attribute record.
(Bug#42238) Cluster API:
Ordered index scans using NdbRecord formerly
expressed a BoundEQ range as separate lower
and upper bounds, resulting in 2 copies of the column values
being sent to the NDB kernel.
Now, when a range is specified by
NdbScanOperation::setBound() , the passed
pointers, key lengths, and inclusive bits are compared, and only
one copy of the equal key columns is sent to the kernel. This
makes such operations more efficient, as half the amount of
KeyInfo is now sent for a
BoundEQ range as before.
(Bug#38793)
Changes in MySQL Cluster NDB 6.3.23 (5.1.32-ndb-6.3.23) Functionality added or changed:
A new data node configuration parameter
MaxLCPStartDelay has been introduced to
facilitate parallel node recovery by causing a local checkpoint
to be delayed while recovering nodes are synchronizing data
dictionaries and other meta-information. For more information
about this parameter, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
(Bug#43053)
Bugs fixed: Performance:
Updates of the SYSTAB_0 system table to
obtain a unique identifier did not use transaction hints for
tables having no primary key. In such cases the NDB kernel used
a cache size of 1. This meant that each insert into a table not
having a primary key required an update of the corresponding
SYSTAB_0 entry, creating a potential
performance bottleneck.
With this fix, inserts on NDB tables without
primary keys can be under some conditions be performed up to
100% faster than previously.
(Bug#39268) Packaging:
Packages for MySQL Cluster were missing the
libndbclient.so and
libndbclient.a files.
(Bug#42278) Partitioning:
Executing ALTER TABLE ... REORGANIZE
PARTITION on an
NDBCLUSTER table having only one
partition caused mysqld to crash.
(Bug#41945) See also Bug#40389.
Backup IDs greater than 231 were not
handled correctly, causing negative values to be used in backup
directory names and printouts.
(Bug#43042)
When using ndbmtd, NDB kernel threads could
hang while trying to start the data nodes with
LockPagesInMainMemory set to 1.
(Bug#43021)
When using multiple management servers and starting several API
nodes (possibly including one or more SQL nodes) whose
connectstrings listed the management servers in different order,
it was possible for 2 API nodes to be assigned the same node ID.
When this happened it was possible for an API node not to get
fully connected, consequently producing a number of errors whose
cause was not easily recognizable.
(Bug#42973)
ndb_error_reporter worked correctly only with
GNU tar. (With other versions of
tar, it produced empty archives.)
(Bug#42753)
Triggers on NDBCLUSTER tables
caused such tables to become locked.
(Bug#42751) See also Bug#16229, Bug#18135.
Given a MySQL Cluster containing no data (that is, whose data
nodes had all been started using --initial , and
into which no data had yet been imported) and having an empty
backup directory, executing START BACKUP with
a user-specified backup ID caused the data nodes to crash.
(Bug#41031)
In some cases, NDB did not check
correctly whether tables had changed before trying to use the
query cache. This could result in a crash of the debug MySQL
server.
(Bug#40464) Disk Data:
It was not possible to add an in-memory column online to a table
that used a table-level or column-level STORAGE
DISK option. The same issue prevented ALTER
ONLINE TABLE ... REORGANIZE PARTITION from working on
Disk Data tables.
(Bug#42549) Disk Data:
Creating a Disk Data tablespace with a very large extent size
caused the data nodes to fail. The issue was observed when using
extent sizes of 100 MB and larger.
(Bug#39096) Disk Data:
Trying to execute a CREATE LOGFILE
GROUP statement using a value greater than
150M for UNDO_BUFFER_SIZE
caused data nodes to crash.
As a result of this fix, the upper limit for
UNDO_BUFFER_SIZE is now
600M ; attempting to set a higher value now
fails gracefully with an error.
(Bug#34102) See also Bug#36702. Disk Data:
When attempting to create a tablespace that already existed, the
error message returned was Table or index with given
name already exists.
(Bug#32662) Disk Data:
Using a path or filename longer than 128 characters for Disk
Data undo log files and tablespace data files caused a number of
issues, including failures of CREATE
LOGFILE GROUP , ALTER LOGFILE
GROUP , CREATE
TABLESPACE , and ALTER
TABLESPACE statements, as well as crashes of
management nodes and data nodes.
With this fix, the maximum length for path and file names used
for Disk Data undo log files and tablespace data files is now
the same as the maximum for the operating system.
(Bug#31769, Bug#31770, Bug#31772) Disk Data:
Attempting to perform a system restart of the cluster where
there existed a logfile group without and undo log files caused
the data nodes to crash.
Note
While issuing a CREATE LOGFILE
GROUP statement without an ADD
UNDOFILE option fails with an error in the MySQL
server, this situation could arise if an SQL node failed
during the execution of a valid CREATE
LOGFILE GROUP statement; it is also possible to
create a logfile group without any undo log files using the
NDB API.
(Bug#17614) Cluster API:
Some error messages from ndb_mgmd contained
newline (\n ) characters. This could break the
MGM API protocol, which uses the newline as a line separator.
(Bug#43104) Cluster API:
When using an ordered index scan without putting all key columns
in the read mask, this invalid use of the NDB API went
undetected, which resulted in the use of uninitialized memory.
(Bug#42591)
Changes in MySQL Cluster NDB 6.3.22 (5.1.31-ndb-6.3.22) Functionality added or changed:
New options are introduced for ndb_restore
for determining which tables or databases should be restored:
--include-tables and
--include-databases can be used to restore
specific tables or databases.
--exclude-tables and
--exclude-databases can be used to exclude
the specified tables or databases from being restored.
For more information about these options, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
(Bug#40429)
Bugs fixed:
When performing more than 32 index or tuple scans on a single
fragment, the scans could be left hanging. This caused
unnecessary timeouts, and in addition could possibly lead to a
hang of an LCP.
(Bug#42559)
A data node failure that occurred between calls to
NdbIndexScanOperation::readTuples(SF_OrderBy)
and NdbTransaction::Execute() was not
correctly handled; a subsequent call to
nextResult() caused a null pointer to be
deferenced, leading to a segfault in mysqld.
(Bug#42545)
Issuing SHOW GLOBAL STATUS LIKE 'NDB%' before
mysqld had connected to the cluster caused a
segmentation fault.
(Bug#42458)
Data node failures that occurred before all data nodes had
connected to the cluster were not handled correctly, leading to
additional data node failures.
(Bug#42422)
When a cluster backup failed with Error 1304 (Node
node_id1 : Backup request from
node_id2 failed to start), no clear
reason for the failure was provided.
As part of this fix, MySQL Cluster now retries backups in the
event of sequence errors.
(Bug#42354) See also Bug#22698.
Issuing SHOW ENGINE
NDBCLUSTER STATUS on an SQL node before the management
server had connected to the cluster caused
mysqld to crash.
(Bug#42264)
Changes in MySQL Cluster NDB 6.3.21 (5.1.31-ndb-6.3.21) Functionality added or changed: Important Change:
Formerly, when the management server failed to create a
transporter for a data node connection,
net_write_timeout seconds
elapsed before the data node was actually allowed to disconnect.
Now in such cases the disconnection occurs immediately.
(Bug#41965) See also Bug#41713.
It is now possible while in Single User Mode to restart all data
nodes using ALL RESTART in the management
client. Restarting of individual nodes while in Single User Mode
remains disallowed.
(Bug#31056)
Formerly, when using MySQL Cluster Replication, records for
“empty” epochs — that is, epochs in which no
changes to NDBCLUSTER data or
tables took place — were inserted into the
ndb_apply_status and
ndb_binlog_index tables on the slave even
when --log-slave-updates was
disabled. Beginning with MySQL Cluster NDB 6.2.16 and MySQL
Cluster NDB 6.3.13 this was changed so that these
“empty” eopchs were no longer logged. However, it
is now possible to re-enable the older behavior (and cause
“empty” epochs to be logged) by using the
--ndb-log-empty-epochs option. For more
information, see Section 16.1.3.3, “Replication Slave Options and Variables”.
See also Bug#37472.
Bugs fixed:
A maximum of 11 TUP scans were allowed in
parallel.
(Bug#42084)
Trying to execute an
ALTER ONLINE TABLE
... ADD COLUMN statement while inserting rows into the
table caused mysqld to crash.
(Bug#41905)
If the master node failed during a global checkpoint, it was
possible in some circumstances for the new master to use an
incorrect value for the global checkpoint index. This could
occur only when the cluster used more than one node group.
(Bug#41469)
API nodes disconnected too agressively from cluster when data
nodes were being restarted. This could sometimes lead to the API
node being unable to access the cluster at all during a rolling
restart.
(Bug#41462)
It was not possible to perform online upgrades from a MySQL
Cluster NDB 6.2 release to MySQL Cluster NDB 6.3.8 or a later
MySQL Cluster NDB 6.3 release.
(Bug#41435)
Cluster log files were opened twice by internal log-handling
code, resulting in a resource leak.
(Bug#41362)
An abort path in the DBLQH kernel block
failed to release a commit acknowledgement marker. This meant
that, during node failure handling, the local query handler
could be added multiple times to the marker record which could
lead to additional node failures due an array overflow.
(Bug#41296)
During node failure handling (of a data node other than the
master), there was a chance that the master was waiting for a
GCP_NODEFINISHED signal from the failed node
after having received it from all other data nodes. If this
occurred while the failed node had a transaction that was still
being committed in the current epoch, the master node could
crash in the DBTC kernel block when
discovering that a transaction actually belonged to an epoch
which was already completed.
(Bug#41295)
Issuing EXIT in the management client
sometimes caused the client to hang.
(Bug#40922)
In the event that a MySQL Cluster backup failed due to file
permissions issues, conflicting reports were issued in the
management client.
(Bug#34526)
If all data nodes were shut down, MySQL clients were unable to
access NDBCLUSTER tables and data
even after the data nodes were restarted, unless the MySQL
clients themselves were restarted.
(Bug#33626) Disk Data:
Starting a cluster under load such that Disk Data tables used
most of the undo buffer could cause data node failures.
The fix for this bug also corrected an issue in the
LGMAN kernel block where the amount of free
space left in the undo buffer was miscalculated, causing buffer
overruns. This could cause records in the buffer to be
overwritten, leading to problems when restarting data nodes.
(Bug#28077) Cluster API:
mgmapi.h contained constructs which only
worked in C++, but not in C.
(Bug#27004)
Changes in MySQL Cluster NDB 6.3.20 (5.1.30-ndb-6.3.20) Bugs fixed:
If a transaction was aborted during the handling of a data node
failure, this could lead to the later handling of an API node
failure not being completed.
(Bug#41214)
Issuing SHOW TABLES repeatedly could cause
NDBCLUSTER tables to be dropped.
(Bug#40854)
Statements of the form UPDATE ... ORDER BY ...
LIMIT run against
NDBCLUSTER tables failed to update
all matching rows, or failed with the error Can't
find record in
'table_name '.
(Bug#40081)
Start phase reporting was inconsistent between the management
client and the cluster log.
(Bug#39667)
Status messages shown in the management client when restarting a
management node were inappropriate and misleading. Now, when
restarting a management node, the messages displayed are as
follows, where node_id is the
management node's node ID:
ndb_mgm> node_id RESTART
Shutting down MGM node node_id for restart
Node node_id is being restarted
ndb_mgm>
(Bug#29275) Disk Data:
This improves on a previous fix for this issue that was made in
MySQL Cluster 6.3.8.
(Bug#37116) See also Bug#29186. Cluster API:
When creating a scan using an NdbScanFilter
object, it was possible to specify conditions against a
BIT column, but the correct rows were not
returned when the scan was executed.
As part of this fix, 4 new comparison operators have been
implemented for use with scans on BIT
columns:
COL_AND_MASK_EQ_MASK
COL_AND_MASK_NE_MASK
COL_AND_MASK_EQ_ZERO
COL_AND_MASK_NE_ZERO
For more information about these operators, see
The NdbScanFilter::BinaryCondition Type.
Equivalent methods are now also defined for
NdbInterpretedCode ; for more information, see
NdbInterpretedCode Bitwise Comparison Operations.
(Bug#40535)
Changes in MySQL Cluster NDB 6.3.19 (5.1.29-ndb-6.3.19) Functionality added or changed: Cluster API: Important Change:
MGM API applications exited without raising any errors if the
connection to the management server was lost. The fix for this
issue includes two changes:
The MGM API now provides its own
SIGPIPE handler to catch the
“broken pipe” error that occurs when writing
to a closed or reset socket. This means that MGM API now
behaves the same as NDB API in this regard.
A new function
ndb_mgm_set_ignore_sigpipe() has been
added to the MGM API. This function makes it possible to
bypass the SIGPIPE handler provded by
the MGM API.
(Bug#40498)
When performing an initial start of a data node, fragment log
files were always created sparsely — that is, not all
bytes were written. Now it is possible to override this behavior
using the new InitFragmentLogFiles
configuration parameter.
(Bug#40847)
Bugs fixed: Cluster API:
Failed operations on BLOB and
TEXT columns were not always
reported correctly to the originating SQL node. Such errors were
sometimes reported as being due to timeouts, when the actual
problem was a transporter overload due to insufficient buffer
space.
(Bug#39867, Bug#39879) Cluster API:
Failed operations on BLOB and
TEXT columns were not always
reported correctly to the originating SQL node. Such errors were
sometimes reported as being due to timeouts, when the actual
problem was a transporter overload due to insufficient buffer
space.
(Bug#39867, Bug#39879)
Undo logs and data files were created in 32K increments. Now
these files are created in 512K increments, resulting in shorter
creation times.
(Bug#40815)
Redo log creation was very slow on some platforms, causing MySQL
Cluster to start more slowly than necessary with some
combinations of hardware and operating system. This was due to
all write operations being synchronized to disk while creating a
redo log file. Now this synchronization occurs only after the
redo log has been created.
(Bug#40734)
Transaction failures took longer to handle than was necessary.
When a data node acting as transaction coordinator (TC) failed,
the surviving data nodes did not inform the API node initiating
the transaction of this until the failure had been processed by
all protocols. However, the API node needed only to know about
failure handling by the transaction protocol — that is, it
needed to be informed only about the TC takeover process. Now,
API nodes (including MySQL servers acting as cluster SQL nodes)
are informed as soon as the TC takeover is complete, so that it
can carry on operating more quickly.
(Bug#40697)
It was theoretically possible for stale data to be read from
NDBCLUSTER tables when the
transaction isolation level was set to
ReadCommitted .
(Bug#40543)
The LockExecuteThreadToCPU and
LockMaintThreadsToCPU parameters did not work
on Solaris.
(Bug#40521)
SET SESSION ndb_optimized_node_selection = 1
failed with an invalid warning message.
(Bug#40457)
A restarting data node could fail with an error in the
DBDIH kernel block when a local or global
checkpoint was started or triggered just as the node made a
request for data from another data node.
(Bug#40370)
Restoring a MySQL Cluster from a dump made using
mysqldump failed due to a spurious error:
Can't execute the given command because you have
active locked tables or an active transaction.
(Bug#40346)
O_DIRECT was incorrectly disabled when making
MySQL Cluster backups.
(Bug#40205)
Heavy DDL usage caused the mysqld processes
to hang due to a timeout error (NDB
error code 266).
(Bug#39885)
Executing EXPLAIN
SELECT on an NDBCLUSTER
table could cause mysqld to crash.
(Bug#39872)
Events logged after setting ALL CLUSTERLOG
STATISTICS=15 in the management client did not always
include the node ID of the reporting node.
(Bug#39839)
The MySQL Query Cache did not function correctly with
NDBCLUSTER tables containing
TEXT columns.
(Bug#39295)
A segfault in Logger::Log caused
ndbd to hang indefinitely. This fix improves
on an earlier one for this issue, first made in MySQL Cluster
NDB 6.2.16 and MySQL Cluster NDB 6.3.17.
(Bug#39180) See also Bug#38609.
Memory leaks could occur in handling of strings used for storing
cluster metadata and providing output to users.
(Bug#38662)
A duplicate key or other error raised when inserting into an
NDBCLUSTER table caused the current
transaction to abort, after which any SQL statement other than a
ROLLBACK
failed. With this fix, the
NDBCLUSTER storage engine now
performs an implicit rollback when a transaction is aborted in
this way; it is no longer necessary to issue an explicit
ROLLBACK
statement, and the next statement that is issued automatically
begins a new transaction.
Note
It remains necessary in such cases to retry the complete
transaction, regardless of which statement caused it to be
aborted.
(Bug#32656) See also Bug#47654.
Error messages for NDBCLUSTER error
codes 1224 and 1227 were missing.
(Bug#28496) Disk Data:
Issuing concurrent CREATE TABLESPACE ,
ALTER TABLESPACE , CREATE LOGFILE
GROUP , or ALTER LOGFILE GROUP
statements on separate SQL nodes caused a resource leak that led
to data node crashes when these statements were used again
later.
(Bug#40921) Disk Data:
Disk-based variable-length columns were not always handled like
their memory-based equivalents, which could potentially lead to
a crash of cluster data nodes.
(Bug#39645) Disk Data:
O_SYNC was incorrectly disabled on platforms
that do not support O_DIRECT . This issue was
noted on Solaris but could have affected other platforms not
having O_DIRECT capability.
(Bug#34638) Cluster API:
The MGM API reset error codes on management server handles
before checking them. This meant that calling an MGM API
function with a null handle caused applications to crash.
(Bug#40455) Cluster API:
It was not always possible to access parent objects directly
from NdbBlob ,
NdbOperation , and
NdbScanOperation objects. To alleviate this
problem, a new getNdbOperation() method has
been added to NdbBlob and new
getNdbTransaction() methods have been added to
NdbOperation and
NdbScanOperation . In addition, a const
variant of NdbOperation::getErrorLine() is
now also available.
(Bug#40242) Cluster API:
NdbScanOperation::getBlobHandle() failed when
used with incorrect column names or numbers.
(Bug#40241) Cluster API:
The MGM API function ndb_mgm_listen_event()
ignored bind addresses.
As part of this fix, it is now possible to specify bind
addresses in connectstrings. See
Section 17.3.2.3, “The MySQL Cluster Connectstring”, for more
information.
(Bug#38473) Cluster API:
The NDB API example programs included in MySQL Cluster source
distributions failed to compile.
(Bug#37491) See also Bug#40238.
Changes in MySQL Cluster NDB 6.3.18 (5.1.28-ndb-6.3.18) Functionality added or changed: Bugs fixed:
When a transaction included a multi-row insert to an
NDBCLUSTER table that caused a
constraint violation, the transaction failed to roll back.
(Bug#395638)
Starting the MySQL Server with the
--ndbcluster option plus an
invalid command-line option (for example, using
mysqld --ndbcluster
--foobar ) caused it to hang while shutting down the
binlog thread.
(Bug#39635)
Dropping and then re-creating a database on one SQL node caused
other SQL nodes to hang.
(Bug#39613)
Setting a low value of MaxNoOfLocalScans
(< 100) and performing a large number of (certain) scans
could cause the Transaction Coordinator to run out of scan
fragment records, and then crash. Now when this resource is
exhausted, the cluster returns Error 291 (Out of
scanfrag records in TC (increase MaxNoOfLocalScans))
instead.
(Bug#39549)
Creating a unique index on an
NDBCLUSTER table caused a memory
leak in the NDB subscription
manager (SUMA ) which could lead to mysqld
hanging, due to the fact that the resource shortage was not
reported back to the NDB kernel
correctly.
(Bug#39518) See also Bug#39450.
Embedded libmysqld with
NDB did not drop table events.
(Bug#39450)
Unique identifiers in tables having no primary key were not
cached. This fix has been observed to increase the efficiency of
INSERT operations on such tables
by as much as 50%.
(Bug#39267)
When restarting a data node, an excessively long shutodwn
message could cause the node process to crash.
(Bug#38580)
After a forced shutdown and initial restart of the cluster, it
was possible for SQL nodes to retain .frm
files corresponding to NDBCLUSTER
tables that had been dropped, and thus to be unaware that these
tables no longer existed. In such cases, attempting to re-create
the tables using CREATE TABLE IF NOT EXISTS
could fail with a spurious Table ... doesn't
exist error.
(Bug#37921)
A statement of the form DELETE FROM
table WHERE
primary_key =value
or UPDATE table WHERE
primary_key =value
where there was no row whose primary key column had the stated
value appeared to succeed, with the
server reporting that 1 row had been changed.
This issue was only known to affect MySQL Cluster NDB 6.3.11 and
later NDB 6.3 versions.
(Bug#37153) Cluster API:
Passing a value greater than 65535 to
NdbInterpretedCode::add_val() and
NdbInterpretedCode::sub_val() caused these
methods to have no effect.
(Bug#39536)
Changes in MySQL Cluster NDB 6.3.17 (5.1.27-ndb-6.3.17) Bugs fixed: Packaging:
Support for the InnoDB storage engine was
missing from the GPL source releases. An updated GPL source
tarball
mysql-5.1.27-ndb-6.3.17-innodb.tar.gz which
includes code for building InnoDB can be
found on
the
MySQL FTP site.
MgmtSrvr::allocNodeId() left a mutex locked
following an Ambiguity for node if %d
error.
(Bug#39158)
An invalid path specification caused
mysql-test-run.pl to fail.
(Bug#39026)
During transactional coordinator takeover (directly after node
failure), the LQH finding an operation in the
LOG_COMMIT state sent an
LQH_TRANS_CONF signal twice, causing the TC
to fail.
(Bug#38930)
An invalid memory access caused the management server to crash
on Solaris Sparc platforms.
(Bug#38628)
A segfault in Logger::Log caused
ndbd to hang indefinitely.
(Bug#38609)
ndb_mgmd failed to start on older Linux
distributions (2.4 kernels) that did not support e-polling.
(Bug#38592)
ndb_mgmd sometimes performed unnecessary
network I/O with the client. This in combination with other
factors led to long-running threads that were attempting to
write to clients that no longer existed.
(Bug#38563)
ndb_restore failed with a floating point
exception due to a division by zero error when trying to restore
certain data files.
(Bug#38520)
A failed connection to the management server could cause a
resource leak in ndb_mgmd.
(Bug#38424)
Failure to parse configuration parameters could cause a memory
leak in the NDB log parser.
(Bug#38380)
Renaming an NDBCLUSTER table on one
SQL node, caused a trigger on this table to be deleted on
another SQL node.
(Bug#36658)
Attempting to add a UNIQUE INDEX twice to an
NDBCLUSTER table, then deleting
rows from the table could cause the MySQL Server to crash.
(Bug#35599)
ndb_restore failed when a single table was
specified.
(Bug#33801)
GCP_COMMIT did not wait for transaction
takeover during node failure. This could cause
GCP_SAVE_REQ to be executed too early. This
could also cause (very rarely) replication to skip rows.
(Bug#30780) Cluster API:
Support for Multi-Range Read index scans using the old API
(using, for example,
NdbIndexScanOperation::setBound() or
NdbIndexScanOperation::end_of_bound() ) were
dropped in MySQL Cluster NDB 6.2. This functionality is restored
in MySQL Cluster NDB 6.3 beginning with 6.3.17, but remains
unavailable in MySQL Cluster NDB 6.2. Both MySQL Cluster NDB 6.2
and 6.3 support Multi-Range Read scans via the
NdbRecord API.
(Bug#38791) Cluster API:
The NdbScanOperation::readTuples() method
could be called multiple times without error.
(Bug#38717) Cluster API:
Certain Multi-Range Read scans involving IS
NULL and IS NOT NULL comparisons
failed with an error in the NDB
local query handler.
(Bug#38204) Cluster API:
Problems with the public headers prevented
NDB applications from being built
with warnings turned on.
(Bug#38177) Cluster API:
Creating an NdbScanFilter object using an
NdbScanOperation object that had not yet had
its readTuples() method called resulted in a
crash when later attempting to use the
NdbScanFilter .
(Bug#37986) Cluster API:
Executing an NdbRecord interpreted delete
created with an ANYVALUE option caused the
transaction to abort.
(Bug#37672)
Changes in MySQL Cluster NDB 6.3.16 (5.1.24-ndb-6.3.16) Functionality added or changed:
Event buffer lag reports are now written to the cluster log.
(Bug#37427)
Added the --no-binlog option for
ndb_restore. When used, this option prevents
information being written to SQL node binary logs from the
restoration of a cluster backup.
(Bug#30452)
Bugs fixed: Cluster API:
Changing the system time on data nodes could cause MGM API
applications to hang and the data nodes to crash.
(Bug#35607) Cluster API:
Changing the system time on data nodes could cause MGM API
applications to hang and the data nodes to crash.
(Bug#35607)
Failure of a data node could sometimes cause mysqld to crash.
(Bug#37628)
DELETE ... WHERE
unique_index_column =value
deleted the wrong row from the table.
(Bug#37516)
If subscription was terminated while a node was down, the epoch
was not properly acknowledged by that node.
(Bug#37442)
libmysqld failed to wait for the cluster
binlog thread to terminate before exiting.
(Bug#37429)
In rare circumstances, a connection followed by a disconnection
could give rise to a “stale” connection where the
connection still existed but was not seen by the transporter.
(Bug#37338)
Queries against NDBCLUSTER tables
were cached only if autocommit
was in use.
(Bug#36692) Cluster API:
When some operations succeeded and some failed following a call
to NdbTransaction::execute(Commit,
AO_IgnoreOnError) , a race condition could cause
spurious occurrences of NDB API Error 4011 (Internal
error).
(Bug#37158) Cluster API:
Creating a table on an SQL node, then starting an NDB API
application that listened for events from this table, then
dropping the table from an SQL node, prevented data node
restarts.
(Bug#32949, Bug#37279) Cluster API:
A buffer overrun in NdbBlob::setValue()
caused erroneous results on Mac OS X.
(Bug#31284)
Changes in MySQL Cluster NDB 6.3.15 (5.1.24-ndb-6.3.15) Bugs fixed:
In certain rare situations, ndb_size.pl
could fail with the error Can't use string
("value ") as a HASH ref while "strict
refs" in use.
(Bug#43022)
Under some circumstances, a failed CREATE
TABLE could mean that subsequent
CREATE TABLE statements caused
node failures.
(Bug#37092)
A fail attempt to create an NDB
table could in some cases lead to resource leaks or cluster
failures.
(Bug#37072)
Attempting to create a native backup of
NDB tables having a large number of
NULL columns and data could lead to node
failures.
(Bug#37039)
Checking of API node connections was not efficiently handled.
(Bug#36843)
Attempting to delete a nonexistent row from a table containing a
TEXT or
BLOB column within a transaction
caused the transaction to fail.
(Bug#36756) See also Bug#36851.
If the combined total of tables and indexes in the cluster was
greater than 4096, issuing START BACKUP
caused data nodes to fail.
(Bug#36044)
Where column values to be compared in a query were of the
VARCHAR or
VARBINARY types,
NDBCLUSTER passed a value padded to
the full size of the column, which caused unnecessary data to be
sent to the data nodes. This also had the effect of wasting CPU
and network bandwidth, and causing condition pushdown to be
disabled where it could (and should) otherwise have been
applied.
(Bug#35393)
When dropping a table failed for any reason (such as when in
single user mode) then the corresponding
.ndb file was still removed.
Cluster API:
Ordered index scans were not pruned correctly where a
partitioning key was specified with an EQ-bound.
(Bug#36950) Cluster API:
When an insert operation involving
BLOB data was attempted on a row
which already existed, no duplicate key error was correctly
reported and the transaction is incorrectly aborted. In some
cases, the existing row could also become corrupted.
(Bug#36851) See also Bug#26756. Cluster API:
NdbApi.hpp depended on
ndb_global.h , which was not actually
installed, causing the compilation of programs that used
NdbApi.hpp to fail.
(Bug#35853)
Changes in MySQL Cluster NDB 6.3.14 (5.1.24-ndb-6.3.14) Bugs fixed:
SET GLOBAL ndb_extra_logging caused
mysqld to crash.
(Bug#36547)
A race condition caused by a failure in epoll handling could
cause data nodes to fail.
(Bug#36537)
Under certain rare circumstances, the failure of the new master
node while attempting a node takeover would cause takeover
errors to repeat without being resolved.
(Bug#36199, Bug#36246, Bug#36247, Bug#36276)
When more than one SQL node connected to the cluster at the same
time, creation of the mysql.ndb_schema table
failed on one of them with an explicit Table
exists error, which was not necessary.
(Bug#35943)
mysqld failed to start after running
mysql_upgrade.
(Bug#35708)
Notification of a cascading master node failures could sometimes
not be transmitted correctly (that is, transmission of the
NF_COMPLETEREP signal could fail), leading to
transactions hanging and timing out
(NDB error 4012), scans hanging,
and failure of the management server process.
(Bug#32645)
If an API node disconnected and then reconnected during Start
Phase 8, then the connection could be “blocked”
— that is, the QMGR kernel block failed
to detect that the API node was in fact connected to the
cluster, causing issues with the
NDB Subscription Manager
(SUMA ).
NDB error 1427 (Api node
died, when SUB_START_REQ reached node) was
incorrectly classified as a schema error rather than a temporary
error.
Cluster API:
Accesing the debug version of libndbclient
via dlopen() resulted in a segmentation
fault.
(Bug#35927) Cluster API:
Attempting to pass a nonexistent column name to the
equal() and setValue()
methods of NdbOperation caused NDB API
applications to crash. Now the column name is checked, and an
error is returned in the event that the column is not found.
(Bug#33747) Cluster API:
Relocation errors were encountered when trying to compile NDB
API applications on a number of platforms, including 64-bit
Linux. As a result, libmysys ,
libmystrings , and libdbug
have been changed from normal libraries to “noinst”
libtool helper libraries. They are no longer
installed as separate libraries; instead, all necessary symbols
from these are added directly to
libndbclient . This means that NDB API
programs now need to be linked only using
-lndbclient .
(Bug#29791)
Changes in MySQL Cluster NDB 6.3.13 (5.1.24-ndb-6.3.13) Bugs fixed: Important Change:
mysqld_safe now traps Signal 13
(SIGPIPE ) so that this signal no longer kills
the MySQL server process.
(Bug#33984)
Node or system restarts could fail due an unitialized variable
in the DTUP kernel block. This issue was
found in MySQL Cluster NDB 6.3.11.
(Bug#35797)
If an error occured while executing a statement involving a
BLOB or
TEXT column of an
NDB table, a memory leak could
result.
(Bug#35593)
It was not possible to determine the value used for the
--ndb-cluster-connection-pool option in the
mysql client. Now this value is reported as a
system status variable.
(Bug#35573)
The ndb_waiter utility wrongly calculated
timeouts.
(Bug#35435)
A SELECT on a table with a
nonindexed, large VARCHAR column
which resulted in condition pushdown on this column could cause
mysqld to crash.
(Bug#35413)
ndb_restore incorrectly handled some
datatypes when applying log files from backups.
(Bug#35343)
In some circumstances, a stopped data node was handled
incorrectly, leading to redo log space being exhausted following
an initial restart of the node, or an initial or partial restart
of the cluster (the wrong CGI might be used in such cases). This
could happen, for example, when a node was stopped following the
creation of a new table, but before a new LCP could be executed.
(Bug#35241)
SELECT ... LIKE ... queries yielded incorrect
results when used on NDB tables. As
part of this fix, condition pushdown of such queries has been
disabled; re-enabling it is expected to be done as part of a
later, permanent fix for this issue.
(Bug#35185)
ndb_mgmd reported errors to
STDOUT rather than to
STDERR .
(Bug#35169)
Nested Multi-Range Read scans failed when the second Multi-Range
Read released the first read's unprocessed operations,
sometimes leading to an SQL node crash.
(Bug#35137)
In some situations, a problem with synchronizing checkpoints
between nodes could cause a system restart or a node restart to
fail with Error 630 during restore of TX.
(Bug#34756)
A node failure during an initial node restart followed by
another node start could cause the master data node to fail,
because it incorrectly gave the node permission to start even if
the invalidated node's LCP was still running.
(Bug#34702)
When a secondary index on a
DECIMAL column was used to
retrieve data from an NDB table, no
results were returned even if the target table had a matched
value in the column that was defined with the secondary index.
(Bug#34515)
An UPDATE on an
NDB table that set a new value for
a unique key column could cause subsequent queries to fail.
(Bug#34208)
If a data node in one node group was placed in the “not
started” state (using
node_id RESTART
-n ), it was not possible to stop a data node in a
different node group.
(Bug#34201)
Numerous NDBCLUSTER test failures
occurred in builds compiled using icc on IA64
platforms.
(Bug#31239)
If a START BACKUP command was issued while
ndb_restore was running, the backup being
restored could be overwritten.
(Bug#26498)
REPLACE statements did not work
correctly with NDBCLUSTER tables
when all columns were not explicitly listed.
(Bug#22045)
CREATE TABLE and
ALTER TABLE statements using
ENGINE=NDB or
ENGINE=NDBCLUSTER caused
mysqld to fail on Solaris 10 for x86
platforms.
(Bug#19911) Cluster API:
Closing a scan before it was executed caused the application to
segfault.
(Bug#36375) Cluster API:
Using NDB API applications from older MySQL Cluster versions
with libndbclient from newer ones caused the
cluster to fail.
(Bug#36124) Cluster API:
Some ordered index scans could return tuples out of order.
(Bug#35908) Cluster API:
Scans having no bounds set were handled incorrectly.
(Bug#35876) Cluster API:
NdbScanFilter::getNdbOperation() , which was
inadvertently removed in MySQL Cluster NDB 6.3.11, has been
restored.
(Bug#35854)
Changes in MySQL Cluster NDB 6.3.10 (5.1.23-ndb-6.3.10) Bugs fixed:
Due to the reduction of the number of local checkpoints from 3
to 2 in MySQL Cluster NDB 6.3.8, a data node using ndbd from
MySQL Cluster NDB 6.3.8 or later started using a file system
from an earlier version could incorrectly invalidate local
checkpoints too early during the startup process, causing the
node to fail.
(Bug#34596)
Changes in MySQL Cluster NDB 6.3.9 (5.1.23-ndb-6.3.9) Bugs fixed:
Cluster failures could sometimes occur when performing more than
three parallel takeovers during node restarts or system
restarts. This affected MySQL Cluster NDB
6.3.x releases only.
(Bug#34445)
Upgrades of a cluster using while a
DataMemory setting in excess of 16 GB caused
data nodes to fail.
(Bug#34378)
Performing many SQL statements on
NDB tables while in
autocommit mode caused a memory
leak in mysqld.
(Bug#34275)
In certain rare circumstances, a race condition could occur
between an aborted insert and a delete leading a data node
crash.
(Bug#34260)
Multi-table updates using ordered indexes during handling of
node failures could cause other data nodes to fail.
(Bug#34216)
When configured with NDB support,
MySQL failed to compile using gcc 4.3 on
64bit FreeBSD systems.
(Bug#34169)
The failure of a DDL statement could sometimes lead to node
failures when attempting to execute subsequent DDL statements.
(Bug#34160)
Extremely long SELECT statements
(where the text of the statement was in excess of 50000
characters) against NDB tables
returned empty results.
(Bug#34107)
When configured with NDB support,
MySQL failed to compile on 64bit FreeBSD systems.
(Bug#34046)
Statements executing multiple inserts performed poorly on
NDB tables having
AUTO_INCREMENT columns.
(Bug#33534)
The ndb_waiter utility polled
ndb_mgmd excessively when obtaining the
status of cluster data nodes.
(Bug#32025) See also Bug#32023.
Transaction atomicity was sometimes not preserved between reads
and inserts under high loads.
(Bug#31477)
Having tables with a great many columns could cause Cluster
backups to fail.
(Bug#30172) Cluster Replication: Disk Data:
Statements violating unique keys on Disk Data tables (such as
attempting to insert NULL into a NOT
NULL column) could cause data nodes to fail. When the
statement was executed from the binlog, this could also result
in failure of the slave cluster.
(Bug#34118) Disk Data:
Updating in-memory columns of one or more rows of Disk Data
table, followed by deletion of these rows and re-insertion of
them, caused data node failures.
(Bug#33619)
Changes in MySQL Cluster NDB 6.3.8 (5.1.23-ndb-6.3.8) Functionality added or changed: Cluster API: Important Change:
Because NDB_LE_MemoryUsage.page_size_kb shows
memory page sizes in bytes rather than kilobytes, it has been
renamed to page_size_bytes . The name
page_size_kb is now deprecated and thus
subject to removal in a future release, although it currently
remains supported for reasons of backward compatibility. See
The Ndb_logevent_type Type, for more information
about NDB_LE_MemoryUsage .
(Bug#30271)
ndb_restore now supports basic
attribute promotion; that is, data from a
column of a given type can be restored to a column using a
“larger” type. For example, Cluster backup data
taken from a SMALLINT column can
be restored to a MEDIUMINT ,
INT , or
BIGINT column.
For more information, see
Section 17.4.17, “ndb_restore — Restore a MySQL Cluster Backup”.
Now only 2 local checkpoints are stored, rather than 3 as in
previous MySQL Cluster versions. This lowers disk space
requirements and reduces the size and number of redo log files
needed.
The mysqld option
--ndb-batch-size has been added. This allows
for controlling the size of batches used for running
transactions.
Node recovery can now be done in parallel, rather than
sequentially, which can result in much faster recovery times.
Persistence of NDB tables can now
be controlled using the session variables
ndb_table_temporary and
ndb_table_no_logging .
ndb_table_no_logging causes
NDB tables not to be checkpointed
to disk; ndb_table_temporary does the same,
and in addition, no schema files are created.
OPTIMIZE TABLE can now be
interrupted. This can be done, for example, by killing the SQL
thread performing the OPTIMIZE operation.
Bugs fixed: Disk Data: Important Change:
It is no longer possible on 32-bit systems to issue statements
appearing to create Disk Data log files or data files greater
than 4 GB in size. (Trying to create log files or data files
larger than 4 GB on 32-bit systems led to unrecoverable data
node failures; such statements now fail with
NDB error 1515.)
(Bug#29186) Replication:
The code implementing heartbeats did not check for possible
errors in some circumstances; this kept the dump thread hanging
while waiting for heartbeats loop even though the slave was no
longer connected.
(Bug#33332)
High numbers of insert operations, delete operations, or both
could cause NDB error 899
(Rowid already allocated) to occur
unnecessarily.
(Bug#34033)
A periodic failure to flush the send buffer by the
NDB TCP transporter could cause a
unnecessary delay of 10 ms between operations.
(Bug#34005)
DROP TABLE did not free all data
memory. This bug was observed in MySQL Cluster NDB 6.3.7 only.
(Bug#33802)
A race condition could occur (very rarely) when the release of a
GCI was followed by a data node failure.
(Bug#33793)
Some tuple scans caused the wrong memory page to be accessed,
leading to invalid results. This issue could affect both
in-memory and Disk Data tables.
(Bug#33739)
A failure to initialize an internal variable led to sporadic
crashes during cluster testing.
(Bug#33715)
The server failed to reject properly the creation of an
NDB table having an unindexed
AUTO_INCREMENT column.
(Bug#30417)
Issuing an
INSERT ...
ON DUPLICATE KEY UPDATE concurrently with or following
a TRUNCATE TABLE statement on an
NDB table failed with
NDB error 4350
Transaction already aborted.
(Bug#29851)
The Cluster backup process could not detect when there was no
more disk space and instead continued to run until killed
manually. Now the backup fails with an appropriate error when
disk space is exhausted.
(Bug#28647)
It was possible in config.ini to define
cluster nodes having node IDs greater than the maximum allowed
value.
(Bug#28298)
Under some circumstances, a recovering data node did not use its
own data, instead copying data from another node even when this
was not required. This in effect bypassed the optimized node
recovery protocol and caused recovery times to be unnecessarily
long.
(Bug#26913) Cluster API:
Transactions containing inserts or reads would hang during
NdbTransaction::execute() calls made from NDB
API applications built against a MySQL Cluster version that did
not support micro-GCPs accessing a later version that supported
micro-GCPs. This issue was observed while upgrading from MySQL
Cluster NDB 6.1.23 to MySQL Cluster NDB 6.2.10 when the API
application built against the earlier version attempted to
access a data node already running the later version, even after
disabling micro-GCPs by setting
TimeBetweenEpochs equal to 0.
(Bug#33895) Cluster API:
When reading a BIT(64) value using
NdbOperation:getValue() , 12 bytes were
written to the buffer rather than the expected 8 bytes.
(Bug#33750)
Changes in MySQL Cluster NDB 6.3.7 (5.1.23-ndb-6.3.7) Functionality added or changed:
Compressed local checkpoints and backups are now supported,
resulting in a space savings of 50% or more over uncompressed
LCPs and backups. Compression of these can be enabled in the
config.ini file using the two new data node
configuration parameters CompressedLCP and
CompressedBackup , respectively.
OPTIMIZE TABLE is now supported
for NDBCLUSTER tables, subject to
the following limitations:
Only in-memory tables are supported.
OPTIMIZE still has no effect on Disk Data
tables.
Only variable-length columns are supported. However, you can
force columns defined using fixed-length data types to be
dynamic using the ROW_FORMAT or
COLUMN_FORMAT option with a
CREATE TABLE or
ALTER TABLE statement.
Memory reclaimed from an NDB table
using OPTIMIZE is generally available to the
cluster, and not confined to the table from which it was
recovered, unlike the case with memory freed using
DELETE .
The performance of OPTIMIZE on
NDB tables can be regulated by
adjusting the value of the
ndb_optimization_delay system variable.
It is now possible to cause statements occurring within the same
transaction to be run as a batch by setting the session variable
transaction_allow_batching to
1 or ON .
Note
To use this feature,
autocommit must be disabled.
Bugs fixed: Partitioning:
When partition pruning on an NDB
table resulted in an ordered index scan spanning only one
partition, any descending flag for the scan was wrongly
discarded, causing ORDER BY DESC to be
treated as ORDER BY ASC ,
MAX() to be handled incorrectly, and similar
problems.
(Bug#33061)
When all data and SQL nodes in the cluster were shut down
abnormally (that is, other than by using STOP
in the cluster management client), ndb_mgm
used excessive amounts of CPU.
(Bug#33237)
When using micro-GCPs, if a node failed while preparing for a
global checkpoint, the master node would use the wrong GCI.
(Bug#32922)
Under some conditions, performing an ALTER
TABLE on an NDBCLUSTER
table failed with a Table is full error,
even when only 25% of DataMemory was in use
and the result should have been a table using less memory (for
example, changing a VARCHAR(100) column to
VARCHAR(80) ).
(Bug#32670)
Changes in MySQL Cluster NDB 6.3.6 (5.1.22-ndb-6.3.6) Functionality added or changed:
Unnecessary reads when performing a primary key or unique key
update have been reduced, and in some cases, eliminated. (It is
almost never necessary to read a record prior to an update, the
lone exception to this being when a primary key is updated,
since this requires a delete followed by an insert, which must
be prepared by reading the record.) Depending on the number of
primary key and unique key lookups that are performed per
transaction, this can yield a considerable improvement in
performance.
Batched operations are now better supported for
DELETE and
UPDATE . (UPDATE
WHERE... and muliple
DELETE .)
Introduced the
Ndb_execute_count status
variable, which measures the number of round trips made by
queries to the NDB kernel.
Bugs fixed:
An insert or update with combined range and equality constraints
failed when run against an NDB
table with the error Got unknown error from
NDB. An example of such a statement would be
UPDATE t1 SET b = 5 WHERE a IN (7,8) OR a >=
10; .
(Bug#31874)
An error with an if statement in
sql/ha_ndbcluster.cc could potentially lead
to an infinite loop in case of failure when working with
AUTO_INCREMENT columns in
NDB tables.
(Bug#31810)
The NDB storage engine code was not
safe for strict-alias optimization in gcc
4.2.1.
(Bug#31761)
ndb_restore displayed incorrect backup file
version information. This meant (for example) that, when
attempting to restore a backup made from a MySQL 5.1.22 cluster
to a MySQL Cluster NDB 6.3.3 cluster, the restore process failed
with the error Restore program older than backup
version. Not supported. Use new restore program.
(Bug#31723)
Following an upgrade, ndb_mgmd would fail
with an ArbitrationError.
(Bug#31690)
The NDB management client command
node_id REPORT
MEMORY provided no output when
node_id was the node ID of a
management or API node. Now, when this occurs, the management
client responds with Node
node_id : is not a data
node .
(Bug#29485)
Performing DELETE operations
after a data node had been shut down could lead to inconsistent
data following a restart of the node.
(Bug#26450)
UPDATE IGNORE could sometimes fail on
NDB tables due to the use of
unitialized data when checking for duplicate keys to be ignored.
(Bug#25817)
Changes in MySQL Cluster NDB 6.3.5 (5.1.22-ndb-6.3.5) Bugs fixed:
A query against a table with TEXT
or BLOB columns that would return
more than a certain amount of data failed with Got
error 4350 'Transaction already aborted' from
NDBCLUSTER.
(Bug#31482) This regression was introduced by Bug#29102.
Changes in MySQL Cluster NDB 6.3.4 (5.1.22-ndb-6.3.4) Functionality added or changed: Incompatible Change:
The --ndb_optimized_node_selection startup
option for mysqld now allows a wider range of
values and corresponding behaviors for SQL nodes when selecting
a transaction coordinator.
You should be aware that the default value and behavior as well
as the value type used for this option have changed, and that
you may need to update the setting used for this option in your
my.cnf file prior to upgrading
mysqld. See
Section 5.1.4, “Server System Variables”, for more information.
Incompatible Change:
The --ndb_optimized_node_selection startup
option for mysqld now allows a wider range of
values and corresponding behaviors for SQL nodes when selecting
a transaction coordinator.
You should be aware that the default value and behavior as well
as the value type used for this option have changed, and that
you may need to update the setting used for this option in your
my.cnf file prior to upgrading
mysqld. See
Section 5.1.4, “Server System Variables”, for more information.
Bugs fixed:
It was possible in some cases for a node group to be
“lost” due to missed local checkpoints following a
system restart.
(Bug#31525)
NDB tables having names containing
nonalphanumeric characters (such as
“$ ”) were not discovered
correctly.
(Bug#31470)
A node failure during a local checkpoint could lead to a
subsequent failure of the cluster during a system restart.
(Bug#31257)
A cluster restart could sometimes fail due to an issue with
table IDs.
(Bug#30975)
Transaction timeouts were not handled well in some
circumstances, leading to excessive number of transactions being
aborted unnecessarily.
(Bug#30379)
In some cases, the cluster managment server logged entries
multiple times following a restart of mgmd.
(Bug#29565)
ndb_mgm --help did not
display any information about the -a option.
(Bug#29509)
An interpreted program of sufficient size and complexity could
cause all cluster data nodes to shut down due to buffer
overruns.
(Bug#29390)
The cluster log was formatted inconsistently and contained
extraneous newline characters.
(Bug#25064)
Changes in MySQL Cluster NDB 6.3.3 (5.1.22-ndb-6.3.3) Functionality added or changed: Bugs fixed: Partitioning:
EXPLAIN
PARTITIONS reported partition usage by queries on
NDB tables according to the
standard MySQL hash function than the hash function used in the
NDB storage engine.
(Bug#29550)
Attempting to restore a backup made on a cluster host using one
endian to a machine using the other endian could cause the
cluster to fail.
(Bug#29674)
The description of the --print option provided
in the output from ndb_restore --help
was incorrect.
(Bug#27683)
Restoring a backup made on a cluster host using one endian to a
machine using the other endian failed for
BLOB and
DATETIME columns.
(Bug#27543, Bug#30024)
Changes in MySQL Cluster NDB 6.3.2 (5.1.22-ndb-6.3.2) Functionality added or changed:
Online ADD COLUMN , ADD
INDEX , and DROP INDEX
operations can now be performed explicitly for
NDB tables, as well as online
renaming of tables and columns for
NDB and MyISAM
tables — that is, without copying or locking of the
affected tables — using ALTER ONLINE
TABLE .
Indexes can also be created and dropped online using
CREATE INDEX and
DROP INDEX , respectively, using
the ONLINE keyword.
You can force operations that would otherwise be performed
online to be done offline using the OFFLINE
keyword.
See Section 12.1.7, “ALTER TABLE Syntax”,
Section 12.1.13, “CREATE INDEX Syntax”, and
Section 12.1.24, “DROP INDEX Syntax”, for more information.
It is now possible to control whether fixed-width or
variable-width storage is used for a given column of an
NDB table by means of the
COLUMN_FORMAT specifier as part of the
column's definition in a CREATE
TABLE or ALTER TABLE
statement.
It is also possible to control whether a given column of an
NDB table is stored in memory or on
disk, using the STORAGE specifier as part of
the column's definition in a CREATE
TABLE or ALTER TABLE
statement.
For permitted values and other information about
COLUMN_FORMAT and STORAGE ,
see Section 12.1.17, “CREATE TABLE Syntax”.
A new cluster management server startup option
--bind-address makes it possible
to restrict management client connections to
ndb_mgmd to a single host and port. For more
information, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
Bugs fixed:
When an NDB event was left behind
but the corresponding table was later recreated and received a
new table ID, the event could not be dropped.
(Bug#30877)
When creating an NDB table with a column that
has COLUMN_FORMAT = DYNAMIC , but the table
tiself uses ROW_FORMAT=FIXED , the table is
considered dynamic, but any columns for which the row format is
unspecified default to FIXED . Now in such
cases the server issues the warning Row format FIXED
incompatible with dynamic attribute
column_name .
(Bug#30276)
An insufficiently descriptive and potentially misleading Error
4006 (Connect failure - out of connection
objects...) was produced when either of the
following two conditions occurred:
There were no more transaction records in the transaction
coordinator
An NDB object in the NDB API
was initialized with insufficient parallelism
Separate error messages are now generated for each of these two
cases.
(Bug#11313)
Changes in MySQL Cluster NDB 6.3.0 (5.1.19-ndb-6.3.0) Functionality added or changed:
Reporting functionality has been significantly enhanced in this
release:
A new configuration parameter
BackupReportFrequency now makes it
possible to cause the management client to provide status
reports at regular intervals as well as for such reports
to be written to the cluster log (depending on cluster
event logging levels). See
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for more
information about this parameter.
A new REPORT command has been added in
the cluster management client. REPORT
BackupStatus allows you to obtain a backup
status report at any time during a backup. REPORT
MemoryUsage reports the current data memory and
index memory used by each data node. For more about the
REPORT command, see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”.
ndb_restore now provides running
reports of its progress when restoring a backup. In
addition, a complete report status report on the backup is
written to the cluster log.
A new configuration parameter ODirect causes
NDB to attempt using
O_DIRECT writes for LCP, backups, and redo
logs, often lowering CPU usage.
17.7.6.3. Changes in the MySQL Cluster NDB 6.2 Series
This section contains unified change history highlights for all
MySQL Cluster releases based on version 6.2 of the
NDBCLUSTER storage engine through
MySQL Cluster NDB 6.2.19. Included are all
changelog entries in the categories MySQL
Cluster, Disk Data, and
Cluster API.
For an overview of features that were added in MySQL Cluster NDB
6.2, see
Section 17.1.4.3, “MySQL CLuster Development in MySQL Cluster NDB 6.2”.
Changes in MySQL Cluster NDB 6.2.18 (5.1.34-ndb-6.2.18) Bugs fixed: Important Change: Partitioning:
User-defined partitioning of an
NDBCLUSTER table without any
primary key sometimes failed, and could cause
mysqld to crash.
Now, if you wish to create an
NDBCLUSTER table with user-defined
partitioning, the table must have an explicit primary key, and
all columns listed in the partitioning expression must be part
of the primary key. The hidden primary key used by the
NDBCLUSTER storage engine is not
sufficient for this purpose. However, if the list of columns is
empty (that is, the table is defined using PARTITION BY
[LINEAR] KEY() ), then no explicit primary key is
required.
This change does not effect the partitioning of tables using any
storage engine other than
NDBCLUSTER .
(Bug#40709)
An internal NDB API buffer was not properly initialized.
(Bug#44977)
When a data node had written its GCI marker to the first page of
a megabyte, and that node was later killed during restart after
having processed that page (marker) but before completing a LCP,
the data node could fail with filesystem errors.
(Bug#44952) See also Bug#42564, Bug#44291.
Inspection of the code revealed that several assignment
operators (= ) were used in place of
comparison operators (== ) in
DbdihMain.cpp .
(Bug#44567) See also Bug#44570.
It was possible for NDB API applications to insert corrupt data
into the database, which could subquently lead to data node
crashes. Now, stricter checking is enforced on input data for
inserts and updates.
(Bug#44132)
TransactionDeadlockDetectionTimeout values
less than 100 were treated as 100. This could cause scans to
time out unexpectedly.
(Bug#44099)
The file ndberror.c contained a C++-style
comment, which caused builds to fail with some C compilers.
(Bug#44036)
A race condition could occur when a data node failed to restart
just before being included in the next global checkpoint. This
could cause other data nodes to fail.
(Bug#43888)
When trying to use a data node with an older version of the
management server, the data node crashed on startup.
(Bug#43699)
Using indexes containing variable-sized columns could lead to
internal errors when the indexes were being built.
(Bug#43226)
In some cases, data node restarts during a system restart could
fail due to insufficient redo log space.
(Bug#43156)
Some queries using combinations of logical and comparison
operators on an indexed column in the WHERE
clause could fail with the error Got error 4541
'IndexBound has no bound information' from
NDBCLUSTER.
(Bug#42857)
ndb_restore --print_data did
not handle DECIMAL columns
correctly.
(Bug#37171)
The output of ndbd --help
did not provide clear information about the program's
--initial and --initial-start
options.
(Bug#28905)
It was theoretically possible for the value of a nonexistent
column to be read as NULL , rather than
causing an error.
(Bug#27843)
When aborting an operation involving both an insert and a
delete, the insert and delete were aborted separately. This was
because the transaction coordinator did not know that the
operations affected on same row, and, in the case of a
committed-read (tuple or index) scan, the abort of the insert
was performed first, then the row was examined after the insert
was aborted but before the delete was aborted. In some cases,
this would leave the row in a inconsistent state. This could
occur when a local checkpoint was performed during a backup.
This issue did not affect primary ley operations or scans that
used locks (these are serialized).
After this fix, for ordered indexes, all operations that follow
the operation to be aborted are now also aborted.
Disk Data: Partitioning:
An NDBCLUSTER table created with a
very large value for the MAX_ROWS option
could — if this table was dropped and a new table with
fewer partitions, but having the same table ID, was created
— cause ndbd to crash when performing a
system restart. This was because the server attempted to examine
each partition whether or not it actually existed.
(Bug#45154) Disk Data:
During a checkpoint, restore points are created for both the
on-disk and in-memory parts of a Disk Data table. Under certain
rare conditions, the in-memory restore point could include or
exclude a row that should have been in the snapshot. This would
later lead to a crash during or following recovery.
(Bug#41915) See also Bug#47832. Disk Data:
When a log file group had an undo log file whose size was too
small, restarting data nodes failed with Read
underflow errors.
As a result of this fix, the minimum allowed
INTIAL_SIZE for an undo log file is now
1M (1 megabyte).
(Bug#29574) Disk Data:
This fix supercedes and improves on an earlier fix made for this
bug in MySQL 5.1.18.
(Bug#24521) Cluster API:
If the largest offset of a
RecordSpecification
used for an
NdbRecord
object was for the NULL bits (and thus not a
column), this offset was not taken into account when calculating
the size used for the RecordSpecification .
This meant that the space for the NULL bits
could be overwritten by key or other information.
(Bug#43891) Cluster API:
The default NdbRecord structures created by
NdbDictionary could have overlapping null
bits and data fields.
(Bug#43590) Cluster API:
When performing insert or write operations,
NdbRecord allows key columns to be specified
in both the key record and in the attribute record. Only one key
column value for each key column should be sent to the NDB
kernel, but this was not guaranteed. This is now ensured as
follows: For insert and write operations, key column values are
taken from the key record; for scan takeover update operations,
key column values are taken from the attribute record.
(Bug#42238) Cluster API:
Ordered index scans using NdbRecord formerly
expressed a BoundEQ range as separate lower
and upper bounds, resulting in 2 copies of the column values
being sent to the NDB kernel.
Now, when a range is specified by
NdbScanOperation::setBound() , the passed
pointers, key lengths, and inclusive bits are compared, and only
one copy of the equal key columns is sent to the kernel. This
makes such operations more efficient, as half the amount of
KeyInfo is now sent for a
BoundEQ range as before.
(Bug#38793)
Changes in MySQL Cluster NDB 6.2.17 (5.1.32-ndb-6.2.17) Functionality added or changed: Important Change:
Formerly, when the management server failed to create a
transporter for a data node connection,
net_write_timeout seconds
elapsed before the data node was actually allowed to disconnect.
Now in such cases the disconnection occurs immediately.
(Bug#41965) See also Bug#41713. Disk Data:
It is now possible to specify default locations for Disk Data
data files and undo log files, either together or separately,
using the data node configuration parameters
FileSystemPathDD ,
FileSystemPathDataFiles , and
FileSystemPathUndoFiles . For information
about these configuration parameters, see
Disk
Data filesystem parameters.
It is also now possible to specify a log file group, tablespace,
or both, that is created when the cluster is started, using the
InitialLogFileGroup and
InitialTablespace data node configuration
parameters. For information about these configuration
parameters, see
Disk
Data object creation parameters.
Bugs fixed: Performance:
Updates of the SYSTAB_0 system table to
obtain a unique identifier did not use transaction hints for
tables having no primary key. In such cases the NDB kernel used
a cache size of 1. This meant that each insert into a table not
having a primary key required an update of the corresponding
SYSTAB_0 entry, creating a potential
performance bottleneck.
With this fix, inserts on NDB tables without
primary keys can be under some conditions be performed up to
100% faster than previously.
(Bug#39268) Packaging:
Packages for MySQL Cluster were missing the
libndbclient.so and
libndbclient.a files.
(Bug#42278) Partitioning:
Executing ALTER TABLE ... REORGANIZE
PARTITION on an
NDBCLUSTER table having only one
partition caused mysqld to crash.
(Bug#41945) See also Bug#40389. Cluster API:
Failed operations on BLOB and
TEXT columns were not always
reported correctly to the originating SQL node. Such errors were
sometimes reported as being due to timeouts, when the actual
problem was a transporter overload due to insufficient buffer
space.
(Bug#39867, Bug#39879) Cluster API:
Failed operations on BLOB and
TEXT columns were not always
reported correctly to the originating SQL node. Such errors were
sometimes reported as being due to timeouts, when the actual
problem was a transporter overload due to insufficient buffer
space.
(Bug#39867, Bug#39879)
Backup IDs greater than 231 were not
handled correctly, causing negative values to be used in backup
directory names and printouts.
(Bug#43042)
When using ndbmtd, NDB kernel threads could
hang while trying to start the data nodes with
LockPagesInMainMemory set to 1.
(Bug#43021)
When using multiple management servers and starting several API
nodes (possibly including one or more SQL nodes) whose
connectstrings listed the management servers in different order,
it was possible for 2 API nodes to be assigned the same node ID.
When this happened it was possible for an API node not to get
fully connected, consequently producing a number of errors whose
cause was not easily recognizable.
(Bug#42973)
ndb_error_reporter worked correctly only with
GNU tar. (With other versions of
tar, it produced empty archives.)
(Bug#42753)
Triggers on NDBCLUSTER tables
caused such tables to become locked.
(Bug#42751) See also Bug#16229, Bug#18135.
When performing more than 32 index or tuple scans on a single
fragment, the scans could be left hanging. This caused
unnecessary timeouts, and in addition could possibly lead to a
hang of an LCP.
(Bug#42559)
A data node failure that occurred between calls to
NdbIndexScanOperation::readTuples(SF_OrderBy)
and NdbTransaction::Execute() was not
correctly handled; a subsequent call to
nextResult() caused a null pointer to be
deferenced, leading to a segfault in mysqld.
(Bug#42545)
Issuing SHOW GLOBAL STATUS LIKE 'NDB%' before
mysqld had connected to the cluster caused a
segmentation fault.
(Bug#42458)
Data node failures that occurred before all data nodes had
connected to the cluster were not handled correctly, leading to
additional data node failures.
(Bug#42422)
When a cluster backup failed with Error 1304 (Node
node_id1 : Backup request from
node_id2 failed to start), no clear
reason for the failure was provided.
As part of this fix, MySQL Cluster now retries backups in the
event of sequence errors.
(Bug#42354) See also Bug#22698.
Issuing SHOW ENGINE
NDBCLUSTER STATUS on an SQL node before the management
server had connected to the cluster caused
mysqld to crash.
(Bug#42264)
A maximum of 11 TUP scans were allowed in
parallel.
(Bug#42084)
Trying to execute an
ALTER ONLINE TABLE
... ADD COLUMN statement while inserting rows into the
table caused mysqld to crash.
(Bug#41905)
If the master node failed during a global checkpoint, it was
possible in some circumstances for the new master to use an
incorrect value for the global checkpoint index. This could
occur only when the cluster used more than one node group.
(Bug#41469)
API nodes disconnected too agressively from cluster when data
nodes were being restarted. This could sometimes lead to the API
node being unable to access the cluster at all during a rolling
restart.
(Bug#41462)
An abort path in the DBLQH kernel block
failed to release a commit acknowledgement marker. This meant
that, during node failure handling, the local query handler
could be added multiple times to the marker record which could
lead to additional node failures due an array overflow.
(Bug#41296)
During node failure handling (of a data node other than the
master), there was a chance that the master was waiting for a
GCP_NODEFINISHED signal from the failed node
after having received it from all other data nodes. If this
occurred while the failed node had a transaction that was still
being committed in the current epoch, the master node could
crash in the DBTC kernel block when
discovering that a transaction actually belonged to an epoch
which was already completed.
(Bug#41295)
If a transaction was aborted during the handling of a data node
failure, this could lead to the later handling of an API node
failure not being completed.
(Bug#41214)
Given a MySQL Cluster containing no data (that is, whose data
nodes had all been started using --initial , and
into which no data had yet been imported) and having an empty
backup directory, executing START BACKUP with
a user-specified backup ID caused the data nodes to crash.
(Bug#41031)
Issuing EXIT in the management client
sometimes caused the client to hang.
(Bug#40922)
Redo log creation was very slow on some platforms, causing MySQL
Cluster to start more slowly than necessary with some
combinations of hardware and operating system. This was due to
all write operations being synchronized to disk while creating a
redo log file. Now this synchronization occurs only after the
redo log has been created.
(Bug#40734)
Transaction failures took longer to handle than was necessary.
When a data node acting as transaction coordinator (TC) failed,
the surviving data nodes did not inform the API node initiating
the transaction of this until the failure had been processed by
all protocols. However, the API node needed only to know about
failure handling by the transaction protocol — that is, it
needed to be informed only about the TC takeover process. Now,
API nodes (including MySQL servers acting as cluster SQL nodes)
are informed as soon as the TC takeover is complete, so that it
can carry on operating more quickly.
(Bug#40697)
It was theoretically possible for stale data to be read from
NDBCLUSTER tables when the
transaction isolation level was set to
ReadCommitted .
(Bug#40543)
In some cases, NDB did not check
correctly whether tables had changed before trying to use the
query cache. This could result in a crash of the debug MySQL
server.
(Bug#40464)
Restoring a MySQL Cluster from a dump made using
mysqldump failed due to a spurious error:
Can't execute the given command because you have
active locked tables or an active transaction.
(Bug#40346)
O_DIRECT was incorrectly disabled when making
MySQL Cluster backups.
(Bug#40205)
Events logged after setting ALL CLUSTERLOG
STATISTICS=15 in the management client did not always
include the node ID of the reporting node.
(Bug#39839)
Start phase reporting was inconsistent between the management
client and the cluster log.
(Bug#39667)
The MySQL Query Cache did not function correctly with
NDBCLUSTER tables containing
TEXT columns.
(Bug#39295)
A segfault in Logger::Log caused
ndbd to hang indefinitely. This fix improves
on an earlier one for this issue, first made in MySQL Cluster
NDB 6.2.16 and MySQL Cluster NDB 6.3.17.
(Bug#39180) See also Bug#38609.
Memory leaks could occur in handling of strings used for storing
cluster metadata and providing output to users.
(Bug#38662)
In the event that a MySQL Cluster backup failed due to file
permissions issues, conflicting reports were issued in the
management client.
(Bug#34526)
A duplicate key or other error raised when inserting into an
NDBCLUSTER table caused the current
transaction to abort, after which any SQL statement other than a
ROLLBACK
failed. With this fix, the
NDBCLUSTER storage engine now
performs an implicit rollback when a transaction is aborted in
this way; it is no longer necessary to issue an explicit
ROLLBACK
statement, and the next statement that is issued automatically
begins a new transaction.
Note
It remains necessary in such cases to retry the complete
transaction, regardless of which statement caused it to be
aborted.
(Bug#32656) See also Bug#47654.
Error messages for NDBCLUSTER error
codes 1224 and 1227 were missing.
(Bug#28496) Disk Data:
It was not possible to add an in-memory column online to a table
that used a table-level or column-level STORAGE
DISK option. The same issue prevented ALTER
ONLINE TABLE ... REORGANIZE PARTITION from working on
Disk Data tables.
(Bug#42549) Disk Data:
Issuing concurrent CREATE TABLESPACE ,
ALTER TABLESPACE , CREATE LOGFILE
GROUP , or ALTER LOGFILE GROUP
statements on separate SQL nodes caused a resource leak that led
to data node crashes when these statements were used again
later.
(Bug#40921) Disk Data:
Disk-based variable-length columns were not always handled like
their memory-based equivalents, which could potentially lead to
a crash of cluster data nodes.
(Bug#39645) Disk Data:
Creating a Disk Data tablespace with a very large extent size
caused the data nodes to fail. The issue was observed when using
extent sizes of 100 MB and larger.
(Bug#39096) Disk Data:
Creation of a tablespace data file whose size was greater than 4
GB failed silently on 32-bit platforms.
(Bug#37116) See also Bug#29186. Disk Data:
O_SYNC was incorrectly disabled on platforms
that do not support O_DIRECT . This issue was
noted on Solaris but could have affected other platforms not
having O_DIRECT capability.
(Bug#34638) Disk Data:
Trying to execute a CREATE LOGFILE
GROUP statement using a value greater than
150M for UNDO_BUFFER_SIZE
caused data nodes to crash.
As a result of this fix, the upper limit for
UNDO_BUFFER_SIZE is now
600M ; attempting to set a higher value now
fails gracefully with an error.
(Bug#34102) See also Bug#36702. Disk Data:
When attempting to create a tablespace that already existed, the
error message returned was Table or index with given
name already exists.
(Bug#32662) Disk Data:
Using a path or filename longer than 128 characters for Disk
Data undo log files and tablespace data files caused a number of
issues, including failures of CREATE
LOGFILE GROUP , ALTER LOGFILE
GROUP , CREATE
TABLESPACE , and ALTER
TABLESPACE statements, as well as crashes of
management nodes and data nodes.
With this fix, the maximum length for path and file names used
for Disk Data undo log files and tablespace data files is now
the same as the maximum for the operating system.
(Bug#31769, Bug#31770, Bug#31772) Disk Data:
Starting a cluster under load such that Disk Data tables used
most of the undo buffer could cause data node failures.
The fix for this bug also corrected an issue in the
LGMAN kernel block where the amount of free
space left in the undo buffer was miscalculated, causing buffer
overruns. This could cause records in the buffer to be
overwritten, leading to problems when restarting data nodes.
(Bug#28077) Disk Data:
Attempting to perform a system restart of the cluster where
there existed a logfile group without and undo log files caused
the data nodes to crash.
Note
While issuing a CREATE LOGFILE
GROUP statement without an ADD
UNDOFILE option fails with an error in the MySQL
server, this situation could arise if an SQL node failed
during the execution of a valid CREATE
LOGFILE GROUP statement; it is also possible to
create a logfile group without any undo log files using the
NDB API.
(Bug#17614) Cluster API:
Some error messages from ndb_mgmd contained
newline (\n ) characters. This could break the
MGM API protocol, which uses the newline as a line separator.
(Bug#43104) Cluster API:
When using an ordered index scan without putting all key columns
in the read mask, this invalid use of the NDB API went
undetected, which resulted in the use of uninitialized memory.
(Bug#42591) Cluster API:
The MGM API reset error codes on management server handles
before checking them. This meant that calling an MGM API
function with a null handle caused applications to crash.
(Bug#40455) Cluster API:
It was not always possible to access parent objects directly
from NdbBlob ,
NdbOperation , and
NdbScanOperation objects. To alleviate this
problem, a new getNdbOperation() method has
been added to NdbBlob and new
getNdbTransaction() methods have been added to
NdbOperation and
NdbScanOperation . In addition, a const
variant of NdbOperation::getErrorLine() is
now also available.
(Bug#40242) Cluster API:
NdbScanOperation::getBlobHandle() failed when
used with incorrect column names or numbers.
(Bug#40241) Cluster API:
The NDB API example programs included in MySQL Cluster source
distributions failed to compile.
(Bug#37491) See also Bug#40238. Cluster API:
mgmapi.h contained constructs which only
worked in C++, but not in C.
(Bug#27004)
Changes in MySQL Cluster NDB 6.2.16 (5.1.28-ndb-6.2.16) Functionality added or changed: Bugs fixed:
Heavy DDL usage caused the mysqld processes
to hang due to a timeout error (NDB
error code 266).
(Bug#39885)
Executing EXPLAIN
SELECT on an NDBCLUSTER
table could cause mysqld to crash.
(Bug#39872)
Starting the MySQL Server with the
--ndbcluster option plus an
invalid command-line option (for example, using
mysqld --ndbcluster
--foobar ) caused it to hang while shutting down the
binlog thread.
(Bug#39635)
Dropping and then re-creating a database on one SQL node caused
other SQL nodes to hang.
(Bug#39613)
Setting a low value of MaxNoOfLocalScans
(< 100) and performing a large number of (certain) scans
could cause the Transaction Coordinator to run out of scan
fragment records, and then crash. Now when this resource is
exhausted, the cluster returns Error 291 (Out of
scanfrag records in TC (increase MaxNoOfLocalScans))
instead.
(Bug#39549)
Creating a unique index on an
NDBCLUSTER table caused a memory
leak in the NDB subscription
manager (SUMA ) which could lead to mysqld
hanging, due to the fact that the resource shortage was not
reported back to the NDB kernel
correctly.
(Bug#39518) See also Bug#39450.
Unique identifiers in tables having no primary key were not
cached. This fix has been observed to increase the efficiency of
INSERT operations on such tables
by as much as 50%.
(Bug#39267)
MgmtSrvr::allocNodeId() left a mutex locked
following an Ambiguity for node if %d
error.
(Bug#39158)
An invalid path specification caused
mysql-test-run.pl to fail.
(Bug#39026)
During transactional coordinator takeover (directly after node
failure), the LQH finding an operation in the
LOG_COMMIT state sent an
LQH_TRANS_CONF signal twice, causing the TC
to fail.
(Bug#38930)
An invalid memory access caused the management server to crash
on Solaris Sparc platforms.
(Bug#38628)
A segfault in Logger::Log caused
ndbd to hang indefinitely.
(Bug#38609)
ndb_mgmd failed to start on older Linux
distributions (2.4 kernels) that did not support e-polling.
(Bug#38592)
When restarting a data node, an excessively long shutodwn
message could cause the node process to crash.
(Bug#38580)
ndb_mgmd sometimes performed unnecessary
network I/O with the client. This in combination with other
factors led to long-running threads that were attempting to
write to clients that no longer existed.
(Bug#38563)
ndb_restore failed with a floating point
exception due to a division by zero error when trying to restore
certain data files.
(Bug#38520)
A failed connection to the management server could cause a
resource leak in ndb_mgmd.
(Bug#38424)
Failure to parse configuration parameters could cause a memory
leak in the NDB log parser.
(Bug#38380)
After a forced shutdown and initial restart of the cluster, it
was possible for SQL nodes to retain .frm
files corresponding to NDBCLUSTER
tables that had been dropped, and thus to be unaware that these
tables no longer existed. In such cases, attempting to re-create
the tables using CREATE TABLE IF NOT EXISTS
could fail with a spurious Table ... doesn't
exist error.
(Bug#37921)
Renaming an NDBCLUSTER table on one
SQL node, caused a trigger on this table to be deleted on
another SQL node.
(Bug#36658)
Attempting to add a UNIQUE INDEX twice to an
NDBCLUSTER table, then deleting
rows from the table could cause the MySQL Server to crash.
(Bug#35599)
ndb_restore failed when a single table was
specified.
(Bug#33801)
GCP_COMMIT did not wait for transaction
takeover during node failure. This could cause
GCP_SAVE_REQ to be executed too early. This
could also cause (very rarely) replication to skip rows.
(Bug#30780) Cluster API:
Passing a value greater than 65535 to
NdbInterpretedCode::add_val() and
NdbInterpretedCode::sub_val() caused these
methods to have no effect.
(Bug#39536) Cluster API:
The NdbScanOperation::readTuples() method
could be called multiple times without error.
(Bug#38717) Cluster API:
Certain Multi-Range Read scans involving IS
NULL and IS NOT NULL comparisons
failed with an error in the NDB
local query handler.
(Bug#38204) Cluster API:
Problems with the public headers prevented
NDB applications from being built
with warnings turned on.
(Bug#38177) Cluster API:
Creating an NdbScanFilter object using an
NdbScanOperation object that had not yet had
its readTuples() method called resulted in a
crash when later attempting to use the
NdbScanFilter .
(Bug#37986) Cluster API:
Executing an NdbRecord interpreted delete
created with an ANYVALUE option caused the
transaction to abort.
(Bug#37672) Cluster API:
Accesing the debug version of libndbclient
via dlopen() resulted in a segmentation
fault.
(Bug#35927)
Changes in MySQL Cluster NDB 6.2.14 (5.1.23-ndb-6.2.14) Functionality added or changed:
Added the MaxBufferedEpochs data node
configuration parameter, which controls the maximum number of
unprocessed epochs by which a subscribing node can lag.
Subscribers which exceed this number are disconnected and forced
to reconnect.
See Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for more
information.
Bugs fixed: Incompatible Change:
The UPDATE statement allowed
NULL to be assigned to NOT
NULL columns (the implicit default value for the
column data type was assigned). This was changed so that on
error occurs.
This change was reverted, because the original report was
determined not to be a bug: Assigning NULL to
a NOT NULL column in an
UPDATE statement should produce
an error only in strict SQL mode and set the column to the
implicit default with a warning otherwise, which was the
original behavior. See Section 10.1.4, “Data Type Default Values”, and
Bug#39265.
(Bug#33699) Cluster API:
Closing a scan before it was executed caused the application to
segfault.
(Bug#36375) Cluster API:
Using NDB API applications from older MySQL Cluster versions
with libndbclient from newer ones caused the
cluster to fail.
(Bug#36124) Cluster API:
Scans having no bounds set were handled incorrectly.
(Bug#35876)
Changes in MySQL Cluster NDB 6.2.13 (5.1.23-ndb-6.2.13) Bugs fixed: Changes in MySQL Cluster NDB 6.2.12 (5.1.23-ndb-6.2.12) Bugs fixed:
Upgrades of a cluster using while a
DataMemory setting in excess of 16 GB caused
data nodes to fail.
(Bug#34378)
Performing many SQL statements on
NDB tables while in
autocommit mode caused a memory
leak in mysqld.
(Bug#34275)
In certain rare circumstances, a race condition could occur
between an aborted insert and a delete leading a data node
crash.
(Bug#34260)
Multi-table updates using ordered indexes during handling of
node failures could cause other data nodes to fail.
(Bug#34216)
When configured with NDB support,
MySQL failed to compile using gcc 4.3 on
64bit FreeBSD systems.
(Bug#34169)
The failure of a DDL statement could sometimes lead to node
failures when attempting to execute subsequent DDL statements.
(Bug#34160)
Extremely long SELECT statements
(where the text of the statement was in excess of 50000
characters) against NDB tables
returned empty results.
(Bug#34107)
Statements executing multiple inserts performed poorly on
NDB tables having
AUTO_INCREMENT columns.
(Bug#33534)
The ndb_waiter utility polled
ndb_mgmd excessively when obtaining the
status of cluster data nodes.
(Bug#32025) See also Bug#32023.
Transaction atomicity was sometimes not preserved between reads
and inserts under high loads.
(Bug#31477)
Having tables with a great many columns could cause Cluster
backups to fail.
(Bug#30172) Cluster Replication: Disk Data:
Statements violating unique keys on Disk Data tables (such as
attempting to insert NULL into a NOT
NULL column) could cause data nodes to fail. When the
statement was executed from the binlog, this could also result
in failure of the slave cluster.
(Bug#34118) Disk Data:
Updating in-memory columns of one or more rows of Disk Data
table, followed by deletion of these rows and re-insertion of
them, caused data node failures.
(Bug#33619)
Changes in MySQL Cluster NDB 6.2.11 (5.1.23-ndb-6.2.11) Functionality added or changed: Cluster API: Important Change:
Because NDB_LE_MemoryUsage.page_size_kb shows
memory page sizes in bytes rather than kilobytes, it has been
renamed to page_size_bytes . The name
page_size_kb is now deprecated and thus
subject to removal in a future release, although it currently
remains supported for reasons of backward compatibility. See
The Ndb_logevent_type Type, for more information
about NDB_LE_MemoryUsage .
(Bug#30271)
Bugs fixed:
High numbers of insert operations, delete operations, or both
could cause NDB error 899
(Rowid already allocated) to occur
unnecessarily.
(Bug#34033)
A periodic failure to flush the send buffer by the
NDB TCP transporter could cause a
unnecessary delay of 10 ms between operations.
(Bug#34005)
A race condition could occur (very rarely) when the release of a
GCI was followed by a data node failure.
(Bug#33793)
Some tuple scans caused the wrong memory page to be accessed,
leading to invalid results. This issue could affect both
in-memory and Disk Data tables.
(Bug#33739)
The server failed to reject properly the creation of an
NDB table having an unindexed
AUTO_INCREMENT column.
(Bug#30417)
Issuing an
INSERT ...
ON DUPLICATE KEY UPDATE concurrently with or following
a TRUNCATE TABLE statement on an
NDB table failed with
NDB error 4350
Transaction already aborted.
(Bug#29851)
The Cluster backup process could not detect when there was no
more disk space and instead continued to run until killed
manually. Now the backup fails with an appropriate error when
disk space is exhausted.
(Bug#28647)
It was possible in config.ini to define
cluster nodes having node IDs greater than the maximum allowed
value.
(Bug#28298) Cluster API:
Transactions containing inserts or reads would hang during
NdbTransaction::execute() calls made from NDB
API applications built against a MySQL Cluster version that did
not support micro-GCPs accessing a later version that supported
micro-GCPs. This issue was observed while upgrading from MySQL
Cluster NDB 6.1.23 to MySQL Cluster NDB 6.2.10 when the API
application built against the earlier version attempted to
access a data node already running the later version, even after
disabling micro-GCPs by setting
TimeBetweenEpochs equal to 0.
(Bug#33895) Cluster API:
When reading a BIT(64) value using
NdbOperation:getValue() , 12 bytes were
written to the buffer rather than the expected 8 bytes.
(Bug#33750)
Changes in MySQL Cluster NDB 6.2.10 (5.1.23-ndb-6.2.10) Bugs fixed: Partitioning:
When partition pruning on an NDB
table resulted in an ordered index scan spanning only one
partition, any descending flag for the scan was wrongly
discarded, causing ORDER BY DESC to be
treated as ORDER BY ASC ,
MAX() to be handled incorrectly, and similar
problems.
(Bug#33061)
When all data and SQL nodes in the cluster were shut down
abnormally (that is, other than by using STOP
in the cluster management client), ndb_mgm
used excessive amounts of CPU.
(Bug#33237)
When using micro-GCPs, if a node failed while preparing for a
global checkpoint, the master node would use the wrong GCI.
(Bug#32922)
Under some conditions, performing an ALTER
TABLE on an NDBCLUSTER
table failed with a Table is full error,
even when only 25% of DataMemory was in use
and the result should have been a table using less memory (for
example, changing a VARCHAR(100) column to
VARCHAR(80) ).
(Bug#32670)
Changes in MySQL Cluster NDB 6.2.9 (5.1.22-ndb-6.2.9) Functionality added or changed: Bugs fixed:
A local checkpoint could sometimes be started before the
previous LCP was restorable from a global checkpoint.
(Bug#32519)
High numbers of API nodes on a slow or congested network could
cause connection negotiation to time out prematurely, leading to
the following issues:
(Bug#32359)
The failure of a master node could lead to subsequent failures
in local checkpointing.
(Bug#32160)
Adding a new TINYTEXT column to
an NDB table which used
COLUMN_FORMAT = DYNAMIC , and when binary
logging was enabled, caused all cluster
mysqld processes to crash.
(Bug#30213)
After adding a new column of one of the
TEXT or
BLOB types to an
NDB table which used
COLUMN_FORMAT = DYNAMIC , it was no longer
possible to access or drop the table using SQL.
(Bug#30205)
A restart of the cluster failed when more than 1 REDO phase was
in use.
(Bug#22696)
Changes in MySQL Cluster NDB 6.2.8 (5.1.22-ndb-6.2.8) Bugs fixed:
In a cluster running in diskless mode and with arbitration
disabled, the failure of a data node during an insert operation
caused other data node to fail.
(Bug#31980)
An insert or update with combined range and equality constraints
failed when run against an NDB
table with the error Got unknown error from
NDB. An example of such a statement would be
UPDATE t1 SET b = 5 WHERE a IN (7,8) OR a >=
10; .
(Bug#31874)
An error with an if statement in
sql/ha_ndbcluster.cc could potentially lead
to an infinite loop in case of failure when working with
AUTO_INCREMENT columns in
NDB tables.
(Bug#31810)
The NDB storage engine code was not
safe for strict-alias optimization in gcc
4.2.1.
(Bug#31761)
Following an upgrade, ndb_mgmd would fail
with an ArbitrationError.
(Bug#31690)
The NDB management client command
node_id REPORT
MEMORY provided no output when
node_id was the node ID of a
management or API node. Now, when this occurs, the management
client responds with Node
node_id : is not a data
node .
(Bug#29485)
Performing DELETE operations
after a data node had been shut down could lead to inconsistent
data following a restart of the node.
(Bug#26450)
UPDATE IGNORE could sometimes fail on
NDB tables due to the use of
unitialized data when checking for duplicate keys to be ignored.
(Bug#25817)
Changes in MySQL Cluster NDB 6.2.7 (5.1.22-ndb-6.2.7) Bugs fixed:
It was possible in some cases for a node group to be
“lost” due to missed local checkpoints following a
system restart.
(Bug#31525)
NDB tables having names containing
nonalphanumeric characters (such as
“$ ”) were not discovered
correctly.
(Bug#31470)
A node failure during a local checkpoint could lead to a
subsequent failure of the cluster during a system restart.
(Bug#31257)
A cluster restart could sometimes fail due to an issue with
table IDs.
(Bug#30975)
Transaction timeouts were not handled well in some
circumstances, leading to excessive number of transactions being
aborted unnecessarily.
(Bug#30379)
In some cases, the cluster managment server logged entries
multiple times following a restart of mgmd.
(Bug#29565)
ndb_mgm --help did not
display any information about the -a option.
(Bug#29509)
The cluster log was formatted inconsistently and contained
extraneous newline characters.
(Bug#25064)
Changes in MySQL Cluster NDB 6.2.6 (5.1.22-ndb-6.2.6) Functionality added or changed: Bugs fixed: Partitioning:
EXPLAIN
PARTITIONS reported partition usage by queries on
NDB tables according to the
standard MySQL hash function than the hash function used in the
NDB storage engine.
(Bug#29550)
When an NDB event was left behind
but the corresponding table was later recreated and received a
new table ID, the event could not be dropped.
(Bug#30877)
Attempting to restore a backup made on a cluster host using one
endian to a machine using the other endian could cause the
cluster to fail.
(Bug#29674)
The description of the --print option provided
in the output from ndb_restore --help
was incorrect.
(Bug#27683)
Restoring a backup made on a cluster host using one endian to a
machine using the other endian failed for
BLOB and
DATETIME columns.
(Bug#27543, Bug#30024)
An insufficiently descriptive and potentially misleading Error
4006 (Connect failure - out of connection
objects...) was produced when either of the
following two conditions occurred:
There were no more transaction records in the transaction
coordinator
An NDB object in the NDB API
was initialized with insufficient parallelism
Separate error messages are now generated for each of these two
cases.
(Bug#11313)
Changes in MySQL Cluster NDB 6.2.5 (5.1.22-ndb-6.2.5) Functionality added or changed:
The following improvements have been made in the
ndb_size.pl utility:
The script can now be used with multiple databases; lists of
databases and tables can also be excluded from analysis.
Schema name information has been added to index table
calculations.
The database name is now an optional parameter, the
exclusion of which causes all databases to be examined.
If selecting from INFORMATION_SCHEMA
fails, the script now attempts to fall back to
SHOW TABLES .
A --real_table_name option has been added;
this designates a table to handle unique index size
calculations.
The report title has been amended to cover cases where more
than one database is being analyzed.
Support for a --socket option was also added.
For more information, see
Section 17.4.21, “ndb_size.pl — NDBCLUSTER Size Requirement Estimator”.
(Bug#28683, Bug#28253)
Online ADD COLUMN , ADD
INDEX , and DROP INDEX
operations can now be performed explicitly for
NDB tables, as well as online
renaming of tables and columns for
NDB and MyISAM
tables — that is, without copying or locking of the
affected tables — using ALTER ONLINE
TABLE .
Indexes can also be created and dropped online using
CREATE INDEX and
DROP INDEX , respectively, using
the ONLINE keyword.
You can force operations that would otherwise be performed
online to be done offline using the OFFLINE
keyword.
See Section 12.1.7, “ALTER TABLE Syntax”,
Section 12.1.13, “CREATE INDEX Syntax”, and
Section 12.1.24, “DROP INDEX Syntax”, for more information.
It is now possible to control whether fixed-width or
variable-width storage is used for a given column of an
NDB table by means of the
COLUMN_FORMAT specifier as part of the
column's definition in a CREATE
TABLE or ALTER TABLE
statement.
It is also possible to control whether a given column of an
NDB table is stored in memory or on
disk, using the STORAGE specifier as part of
the column's definition in a CREATE
TABLE or ALTER TABLE
statement.
For permitted values and other information about
COLUMN_FORMAT and STORAGE ,
see Section 12.1.17, “CREATE TABLE Syntax”.
A new cluster management server startup option
--bind-address makes it possible
to restrict management client connections to
ndb_mgmd to a single host and port. For more
information, see
Section 17.4.4, “ndb_mgmd — The MySQL Cluster Management Server Daemon”.
Bugs fixed:
When handling BLOB columns, the
addition of read locks to the lock queue was not handled
correctly.
(Bug#30764)
Discovery of NDB tables did not
work correctly with INFORMATION_SCHEMA .
(Bug#30667)
A file system close operation could fail during a node or system
restart.
(Bug#30646)
Using the --ndb-cluster-connection-pool option
for mysqld caused DDL statements to be
executed twice.
(Bug#30598)
When creating an NDB table with a column that
has COLUMN_FORMAT = DYNAMIC , but the table
tiself uses ROW_FORMAT=FIXED , the table is
considered dynamic, but any columns for which the row format is
unspecified default to FIXED . Now in such
cases the server issues the warning Row format FIXED
incompatible with dynamic attribute
column_name .
(Bug#30276)
ndb_size.pl failed on tables with
FLOAT columns whose definitions
included commas (for example, FLOAT(6,2) ).
(Bug#29228)
Reads on BLOB columns were not
locked when they needed to be to guarantee consistency.
(Bug#29102) See also Bug#31482.
A query using joins between several large tables and requiring
unique index lookups failed to complete, eventually returning
Uknown Error after a very long period of
time. This occurred due to inadequate handling of instances
where the Transaction Coordinator ran out of
TransactionBufferMemory , when the cluster
should have returned NDB error code 4012 (Request
ndbd time-out).
(Bug#28804)
An attempt to perform a SELECT ... FROM
INFORMATION_SCHEMA.TABLES whose result included
information about NDB tables for
which the user had no privileges crashed the MySQL Server on
which the query was performed.
(Bug#26793) Cluster API:
A call to CHECK_TIMEDOUT_RET() in
mgmapi.cpp should have been a call to
DBUG_CHECK_TIMEDOUT_RET() .
(Bug#30681)
Changes in MySQL Cluster NDB 6.2.4 (5.1.19-ndb-6.2.4) Bugs fixed:
When restarting a data node, queries could hang during that
node's start phase 5, and continue only after the node had
entered phase 6.
(Bug#29364)
Replica redo logs were inconsistently handled during a system
restart.
(Bug#29354) Disk Data:
Performing Disk Data schema operations during a node restart
could cause forced shutdowns of other data nodes.
(Bug#29501) Disk Data:
Disk data meta-information that existed in
ndbd might not be visible to
mysqld.
(Bug#28720) Disk Data:
The number of free extents was incorrectly reported for some
tablespaces.
(Bug#28642)
Changes in MySQL Cluster NDB 6.2.3 (5.1.19-ndb-6.2.3) Functionality added or changed: Important Change:
The TimeBetweenWatchdogCheckInitial
configuration parameter was added to allow setting of a separate
watchdog timeout for memory allocation during startup of the
data nodes. See Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”,
for more information.
(Bug#28899) Cluster API: Important Change:
A new NdbRecord object has been added to the
NDB API. This object provides
mapping to a record stored in NDB .
See The NdbRecord Interface, for more information.
auto_increment_increment and
auto_increment_offset are now
supported for NDB tables.
(Bug#26342)
A REPORT BackupStatus command has been added
in the cluster management client. This command allows you to
obtain a backup status report at any time during a backup. For
more about this command, see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”.
Reporting functionality has been significantly enhanced in this
release:
A new configuration parameter
BackupReportFrequency now makes it
possible to cause the management client to provide status
reports at regular intervals as well as for such reports
to be written to the cluster log (depending on cluster
event logging levels). See
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for more
information about this parameter.
A new REPORT command has been added in
the cluster management client. REPORT
BackupStatus allows you to obtain a backup
status report at any time during a backup. REPORT
MemoryUsage reports the current data memory and
index memory used by each data node. For more about the
REPORT command, see
Section 17.5.2, “Commands in the MySQL Cluster Management Client”.
ndb_restore now provides running
reports of its progress when restoring a backup. In
addition, a complete report status report on the backup is
written to the cluster log.
A new configuration parameter ODirect causes
NDB to attempt using
O_DIRECT writes for LCP, backups, and redo
logs, often lowering CPU usage.
ndb_restore now provides running reports of
its progress when restoring a backup. In addition, a complete
report status report on the backup is written to the cluster
log.
A new configuration parameter
BackupReportFrequency now makes it possible
to cause the management client to provide status reports at
regular intervals as well as for such reports to be written to
the cluster log (depending on cluster event logging levels). See
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”, for more
information about this parameter.
A new memory allocator has been implemented for the
NDB kernel, which allocates memory
to tables 32K page by 32K page rather than allocating it in
variable-sized chunks as previously. This removes much of the
memory overhead that was associated with the old memory
allocator.
Bugs fixed:
When a node failed to respond to a COPY_GCI
signal as part of a global checkpoint, the master node was
killed instead of the node that actually failed.
(Bug#29331)
Memory corruption could occur due to a problem in the
DBTUP kernel block.
(Bug#29229)
A query having a large IN(...) or
NOT IN(...) list in the
WHERE condition on an
NDB table could cause
mysqld to crash.
(Bug#29185)
In the event that two data nodes in the same node group and
participating in a GCP crashed before they had written their
respective P0.sysfile files,
QMGR could refuse to start, issuing an
invalid Insufficient nodes for restart
error instead.
(Bug#29167)
An invalid comparison made during REDO
validation that could lead to an Error while reading
REDO log condition.
(Bug#29118)
Attempting to restore a NULL row to a
VARBINARY column caused
ndb_restore to fail.
(Bug#29103)
ndb_error_reporter now preserves timestamps
on files.
(Bug#29074)
The wrong data pages were sometimes invalidated following a
global checkpoint.
(Bug#29067)
If at least 2 files were involved in REDO
invalidation, then file 0 of page 0 was not updated and so
pointed to an invalid part of the redo log.
(Bug#29057)
It is now possible to set the maximum size of the allocation
unit for table memory using the MaxAllocate
configuration parameter.
(Bug#29044)
When shutting down mysqld, the
NDB binlog process was not shut
down before log cleanup began.
(Bug#28949)
A corrupt schema file could cause a File already
open error.
(Bug#28770)
Having large amounts of memory locked caused swapping to disk.
(Bug#28751)
Setting InitialNoOpenFiles equal to
MaxNoOfOpenFiles caused an error. This was
due to the fact that the actual value of
MaxNoOfOpenFiles as used by the cluster was
offset by 1 from the value set in
config.ini .
(Bug#28749)
LCP files were not removed following an initial system restart.
(Bug#28726)
UPDATE IGNORE statements involving the
primary keys of multiple tables could result in data corruption.
(Bug#28719)
A race condition could result when nonmaster nodes (in addition
to the master node) tried to update active status due to a local
checkpoint (that is, between NODE_FAILREP and
COPY_GCIREQ events). Now only the master
updates the active status.
(Bug#28717)
A fast global checkpoint under high load with high usage of the
redo buffer caused data nodes to fail.
(Bug#28653)
The management client's response to START BACKUP
WAIT COMPLETED did not include the backup ID.
(Bug#27640) Disk Data:
When dropping a page, the stack's bottom entry could sometime be
left “cold” rather than “hot”,
violating the rules for stack pruning.
(Bug#29176) Disk Data:
When loading data into a cluster following a version upgrade,
the data nodes could forcibly shut down due to page and buffer
management failures (that is, ndbrequire
failures in PGMAN ).
(Bug#28525) Disk Data:
Repeated INSERT and
DELETE operations on a Disk Data
table having one or more large
VARCHAR columns could cause data
nodes to fail.
(Bug#20612) Cluster API:
The timeout set using the MGM API
ndb_mgm_set_timeout() function was
incorrectly interpreted as seconds rather than as milliseconds.
(Bug#29063) Cluster API:
An invalid error code could be set on transaction objects by
BLOB handling code.
(Bug#28724)
Changes in MySQL Cluster NDB 6.2.2 (5.1.18-ndb-6.2.2) Functionality added or changed:
New cluster management client DUMP commands
were added to aid in tracking transactions, scan operations, and
locks. See DUMP 2350 ,
DUMP 2352 , and
DUMP 2550 , for more
information.
Added the mysqld option
--ndb-cluster-connection-pool that allows a
single MySQL server to use multiple connections to the cluster.
This allows for scaling out using multiple MySQL clients per SQL
node instead of or in addition to using multiple SQL nodes with
the cluster.
For more information about this option, see
Section 17.3.4, “MySQL Server Options and Variables for MySQL Cluster”.
Changes in MySQL Cluster NDB 6.2.1 (5.1.18-ndb-6.2.1) Bugs fixed:
Multiple operations involving deletes followed by reads were not
handled correctly.
Note
This issue could also affect MySQL Cluster Replication.
(Bug#28276) Cluster API:
Using NdbBlob::writeData() to write data in
the middle of an existing blob value (that is, updating the
value) could overwrite some data past the end of the data to be
changed.
(Bug#27018)
Changes in MySQL Cluster NDB 6.2.0 (5.1.16-ndb-6.2.0) Functionality added or changed: 17.7.6.4. Changes in the MySQL Cluster NDB 6.1 Series
This section contains unified change history highlights for all
MySQL Cluster releases based on version 6.1 of the
NDBCLUSTER storage engine through
MySQL Cluster NDB 5.1.15-ndb-6.1.23. Included are all
changelog entries in the categories MySQL
Cluster, Disk Data, and
Cluster API.
For an overview of features that were added in MySQL Cluster NDB
6.1, see
Section 17.1.4.2, “MySQL Cluster Development in MySQL Cluster NDB 6.1”.
Note
MySQL Cluster NDB 6.1 is no longer being developed or
maintained, and the information presented in this section
should be considered to be of historical interest only. If you
are using MySQL Cluster NDB 6.1, you should upgrade as soon as
possible to the most recent version of MySQL Cluster NDB 6.2
or later MySQL Cluster release series.
Changes in MySQL Cluster NDB 6.1.23 (5.1.15-ndb-6.1.23) Bugs fixed: Changes in MySQL Cluster NDB 6.1.22 (5.1.15-ndb-6.1.22) Bugs fixed: Changes in MySQL Cluster NDB 6.1.21 (5.1.15-ndb-6.1.21) Bugs fixed:
A node failure during a local checkpoint could lead to a
subsequent failure of the cluster during a system restart.
(Bug#31257)
A cluster restart could sometimes fail due to an issue with
table IDs.
(Bug#30975)
Changes in MySQL Cluster NDB 6.1.19 (5.1.15-ndb-6.1.19) Functionality added or changed: Changes in MySQL Cluster NDB 6.1.18 (5.1.15-ndb-6.1.18) Bugs fixed:
When restarting a data node, queries could hang during that
node's start phase 5, and continue only after the node had
entered phase 6.
(Bug#29364) Disk Data:
Disk data meta-information that existed in
ndbd might not be visible to
mysqld.
(Bug#28720) Disk Data:
The number of free extents was incorrectly reported for some
tablespaces.
(Bug#28642)
Changes in MySQL Cluster NDB 6.1.17 (5.1.15-ndb-6.1.17) Bugs fixed: Changes in MySQL Cluster NDB 6.1.16 (5.1.15-ndb-6.1.16) Bugs fixed:
When a node failed to respond to a COPY_GCI
signal as part of a global checkpoint, the master node was
killed instead of the node that actually failed.
(Bug#29331)
An invalid comparison made during REDO
validation that could lead to an Error while reading
REDO log condition.
(Bug#29118)
The wrong data pages were sometimes invalidated following a
global checkpoint.
(Bug#29067)
If at least 2 files were involved in REDO
invalidation, then file 0 of page 0 was not updated and so
pointed to an invalid part of the redo log.
(Bug#29057) Disk Data:
When dropping a page, the stack's bottom entry could sometime be
left “cold” rather than “hot”,
violating the rules for stack pruning.
(Bug#29176)
Changes in MySQL Cluster NDB 6.1.15 (5.1.15-ndb-6.1.15) Bugs fixed: Changes in MySQL Cluster NDB 6.1.14 (5.1.15-ndb-6.1.14) Bugs fixed:
In the event that two data nodes in the same node group and
participating in a GCP crashed before they had written their
respective P0.sysfile files,
QMGR could refuse to start, issuing an
invalid Insufficient nodes for restart
error instead.
(Bug#29167)
Changes in MySQL Cluster NDB 6.1.13 (5.1.15-ndb-6.1.13) Bugs fixed: Changes in MySQL Cluster NDB 6.1.12 (5.1.15-ndb-6.1.12) Functionality added or changed:
New cluster management client DUMP commands
were added to aid in tracking transactions, scan operations, and
locks. See DUMP 2350 ,
DUMP 2352 , and
DUMP 2550 .
Bugs fixed: Changes in MySQL Cluster NDB 6.1.11 (5.1.15-ndb-6.1.11) Functionality added or changed: Important Change:
The TimeBetweenWatchdogCheckInitial
configuration parameter was added to allow setting of a separate
watchdog timeout for memory allocation during startup of the
data nodes. See Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”,
for more information.
(Bug#28899)
A new configuration parameter ODirect causes
NDB to attempt using
O_DIRECT writes for LCP, backups, and redo
logs, often lowering CPU usage.
Bugs fixed:
Having large amounts of memory locked caused swapping to disk.
(Bug#28751)
LCP files were not removed following an initial system restart.
(Bug#28726) Disk Data:
Repeated INSERT and
DELETE operations on a Disk Data
table having one or more large
VARCHAR columns could cause data
nodes to fail.
(Bug#20612)
Changes in MySQL Cluster NDB 6.1.10 (5.1.15-ndb-6.1.10) Functionality added or changed: Bugs fixed:
A regression in the heartbeat monitoring code could lead to node
failure under high load. This issue affected MySQL 5.1.19 and
MySQL Cluster NDB 6.1.10 only.
(Bug#28783)
A corrupt schema file could cause a File already
open error.
(Bug#28770)
Setting InitialNoOpenFiles equal to
MaxNoOfOpenFiles caused an error. This was
due to the fact that the actual value of
MaxNoOfOpenFiles as used by the cluster was
offset by 1 from the value set in
config.ini .
(Bug#28749)
A race condition could result when nonmaster nodes (in addition
to the master node) tried to update active status due to a local
checkpoint (that is, between NODE_FAILREP and
COPY_GCIREQ events). Now only the master
updates the active status.
(Bug#28717)
A fast global checkpoint under high load with high usage of the
redo buffer caused data nodes to fail.
(Bug#28653) Disk Data:
When loading data into a cluster following a version upgrade,
the data nodes could forcibly shut down due to page and buffer
management failures (that is, ndbrequire
failures in PGMAN ).
(Bug#28525)
Changes in MySQL Cluster NDB 6.1.9 (5.1.15-ndb-6.1.9) Bugs fixed:
When an API node sent more than 1024 signals in a single batch,
NDB would process only the first
1024 of these, and then hang.
(Bug#28443) Disk Data:
The cluster backup process scanned in ACC
index order, which had bad effects for disk data.
(Bug#28593)
Changes in MySQL Cluster NDB 6.1.8 (5.1.15-ndb-6.1.8) Bugs fixed:
Local checkpoint files relating to dropped
NDB tables were not removed.
(Bug#28348)
Repeated insertion of data generated by
mysqldump into
NDB tables could eventually lead to
failure of the cluster.
(Bug#27437) Disk Data:
Extremely large inserts into Disk Data tables could lead to data
node failure in some circumstances.
(Bug#27942) Cluster API:
In a multi-operation transaction, a delete operation followed by
the insertion of an implicit NULL failed to
overwrite an existing value.
(Bug#20535)
Changes in MySQL Cluster NDB 6.1.7 (5.1.15-ndb-6.1.7) Functionality added or changed: Cluster Replication: Incompatible Change:
The schema for the ndb_apply_status table in
the mysql system database has changed. When
upgrading to this release from a previous MySQL Cluster NDB 6.x
or mainline MySQL 5.1 release, you must drop the
mysql.ndb_apply_status table, then restart
the server in order for the table to be re-created with the new
schema.
See Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
additional information.
Bugs fixed:
The cluster waited 30 seconds instead of 30 milliseconds before
reading table statistics.
(Bug#28093)
Under certain rare circumstances, ndbd could
get caught in an infinite loop when one transaction took a read
lock and then a second transaction attempted to obtain a write
lock on the same tuple in the lock queue.
(Bug#28073)
Under some circumstances, a node restart could fail to update
the Global Checkpoint Index (GCI).
(Bug#28023)
An INSERT followed by a delete
DELETE on the same
NDB table caused a memory leak.
(Bug#27756) This regression was introduced by Bug#20612.
Under certain rare circumstances performing a
DROP TABLE or
TRUNCATE TABLE on an
NDB table could cause a node
failure or forced cluster shutdown.
(Bug#27581)
Memory usage of a mysqld process grew even
while idle.
(Bug#27560)
Performing a delete followed by an insert during a local
checkpoint could cause a Rowid already
allocated error.
(Bug#27205) Cluster Replication: Disk Data:
An issue with replication of Disk Data tables could in some
cases lead to node failure.
(Bug#28161) Disk Data:
Changes to a Disk Data table made as part of a transaction could
not be seen by the client performing the changes until the
transaction had been committed.
(Bug#27757) Disk Data:
When restarting a data node following the creation of a large
number of Disk Data objects (approximately 200 such objects),
the cluster could not assign a node ID to the restarting node.
(Bug#25741) Disk Data:
Changing a column specification or issuing a
TRUNCATE TABLE statement on a
Disk Data table caused the table to become an in-memory table.
This fix supersedes an incomplete fix that was made for this
issue in MySQL 5.1.15.
(Bug#24667, Bug#25296) Cluster API:
An issue with the way in which the
NdbDictionary::Dictionary::listEvents()
method freed resources could sometimes lead to memory
corruption.
(Bug#27663)
Changes in MySQL Cluster NDB 6.1.6 (5.1.15-ndb-6.1.6) Functionality added or changed: Cluster Replication: Incompatible Change:
The schema for the ndb_apply_status table in
the mysql system database has changed. When
upgrading to this release from a previous MySQL Cluster NDB 6.x
or mainline MySQL 5.1 release, you must drop the
mysql.ndb_apply_status table, then restart
the server in order for the table to be re-created with the new
schema.
See Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
additional information.
Bugs fixed:
A data node failing while another data node was restarting could
leave the cluster in an inconsistent state. In certain rare
cases, this could lead to a race condition and the eventual
forced shutdown of the cluster.
(Bug#27466)
It was not possible to set
LockPagesInMainMemory equal to
0 .
(Bug#27291)
A race condition could sometimes occur if the node acting as
master failed while node IDs were still being allocated during
startup.
(Bug#27286)
When a data node was taking over as the master node, a race
condition could sometimes occur as the node was assuming
responsibility for handling of global checkpoints.
(Bug#27283)
mysqld could crash shortly after a data node
failure following certain DML operations.
(Bug#27169)
The same failed request from an API node could be handled by the
cluster multiple times, resulting in reduced performance.
(Bug#27087)
The failure of a data node while restarting could cause other
data nodes to hang or crash.
(Bug#27003)
mysqld processes would sometimes crash under
high load.
Note
This fix improves on and replaces a fix for this bug that was
made in MySQL Cluster NDB 6.1.5.
(Bug#26825) Disk Data:
DROP INDEX on a Disk Data table
did not always move data from memory into the tablespace.
(Bug#25877) Cluster API:
An issue with the way in which the
NdbDictionary::Dictionary::listEvents()
method freed resources could sometimes lead to memory
corruption.
(Bug#27663) Cluster API:
A delete operation using a scan followed by an insert using a
scan could cause a data node to fail.
(Bug#27203)
Changes in MySQL Cluster NDB 6.1.5 (5.1.15-ndb-6.1.5) Functionality added or changed: Cluster Replication: Incompatible Change:
The schema for the ndb_apply_status table in
the mysql system database has changed. When
upgrading to this release from a previous MySQL Cluster NDB 6.x
or mainline MySQL 5.1 release, you must drop the
mysql.ndb_apply_status table, then restart
the server in order for the table to be re-created with the new
schema.
See Section 17.6.4, “MySQL Cluster Replication Schema and Tables”, for
additional information.
Bugs fixed:
Creating a table on one SQL node while in single user mode
caused other SQL nodes to crash.
(Bug#26997)
mysqld processes would sometimes crash under
high load.
Note
This fix was reverted in MySQL Cluster NDB 6.1.6.
(Bug#26825)
An infinite loop in an internal logging function could cause
trace logs to fill up with Unknown Signal
type error messages and thus grow to unreasonable
sizes.
(Bug#26720) Disk Data:
When creating a log file group, setting
INITIAL_SIZE to less than
UNDO_BUFFER_SIZE caused data nodes to crash.
(Bug#25743)
Changes in MySQL Cluster NDB 6.1.4 (5.1.15-ndb-6.1.4) Functionality added or changed:
An ndb_wait_connected system
variable has been added for mysqld. It causes
mysqld wait a specified amount of time to be
connected to the cluster before accepting client connections.
For more information, see
Section 17.3.4.3, “MySQL Cluster System Variables”.
Cluster API:
It is now possible to specify the transaction coordinator when
starting a transaction. See
Ndb::startTransaction() , for more
information.
Cluster API:
It is now possible to iterate over all existing
NDB objects using three new methods
of the Ndb_cluster_connection class:
lock_ndb_objects()
get_next_ndb_object()
unlock_ndb_objects()
For more information about these methods and their use, see
ndb_cluster_connection::get_next_ndb_object() ,
in the MySQL Cluster API Guide.
Bugs fixed:
Using only the --print_data option (and no
other options) with ndb_restore caused
ndb_restore to fail.
(Bug#26741) This regression was introduced by Bug#14612.
An inadvertent use of unaligned data caused
ndb_restore to fail on some 64-bit platforms,
including Sparc and Itanium-2.
(Bug#26739)
Changes in MySQL Cluster NDB 6.1.3 (5.1.15-ndb-6.1.3) Functionality added or changed: Bugs fixed:
An invalid pointer was returned following a
FSCLOSECONF signal when accessing the REDO
logs during a node restart or system restart.
(Bug#26515)
The InvalidUndoBufferSize error used the
same error code (763) as the
IncompatibleVersions error.
InvalidUndoBufferSize now uses its own
error code (779).
(Bug#26490)
The failure of a data node when restarting it with
--initial could lead to failures of subsequent
data node restarts.
(Bug#26481)
Takeover for local checkpointing due to multiple failures of
master nodes was sometimes incorrectly handled.
(Bug#26457)
The LockPagesInMainMemory parameter was not
read until after distributed communication had already started
between cluster nodes. When the value of this parameter was
1 , this could sometimes result in data node
failure due to missed heartbeats.
(Bug#26454)
Under some circumstances, following the restart of a management
node, all data nodes would connect to it normally, but some of
them subsequently failed to log any events to the management
node.
(Bug#26293)
No appropriate error message was provided when there was
insufficient REDO log file space for the cluster to start.
(Bug#25801)
A memory allocation failure in SUMA (the
cluster Subscription Manager) could cause the cluster to crash.
(Bug#25239)
The message Error 0 in readAutoIncrementValue(): no
Error was written to the error log whenever
SHOW TABLE STATUS was performed
on a Cluster table that did not have an
AUTO_INCREMENT column.
Note
This improves on and supersedes an earlier fix that was made
for this issue in MySQL 5.1.12.
(Bug#21033) Disk Data:
A memory overflow could occur with tables having a large amount
of data stored on disk, or with queries using a very high degree
of parallelism on Disk Data tables.
(Bug#26514) Disk Data:
Use of a tablespace whose INITIAL_SIZE was
greater than 1 GB could cause the cluster to crash.
(Bug#26487)
Changes in MySQL Cluster NDB 6.1.2 (5.1.15-ndb-6.1.2) Bugs fixed: Changes in MySQL Cluster NDB 6.1.1 (5.1.15-ndb-6.1.1) Functionality added or changed: Bugs fixed:
A memory leak could cause problems during a node or cluster
shutdown or failure.
(Bug#25997) Cluster API: Disk Data:
A delete and a read performed in the same operation could cause
one or more data nodes to crash. This could occur when the
operation affected more than 5 columns concurrently, or when one
or more of the columns was of the
VARCHAR type and was stored on
disk.
(Bug#25794) Cluster API: Disk Data:
A delete and a read performed in the same operation could cause
one or more data nodes to crash. This could occur when the
operation affected more than 5 columns concurrently, or when one
or more of the columns was of the
VARCHAR type and was stored on
disk.
(Bug#25794)
Changes in MySQL Cluster NDB 6.1.0 (5.1.14-ndb-6.1.0) Functionality added or changed:
A new configuration parameter
MemReportFrequency allows for additional
control of data node memory usage. Previously, only warnings at
predetermined percentages of memory allocation were given;
setting this parameter allows for that behavior to be
overridden. For more information, see
Section 17.3.2.6, “Defining MySQL Cluster Data Nodes”.
Bugs fixed:
When a data node was shut down using the management client
STOP command, a connection event
(NDB_LE_Connected ) was logged instead of a
disconnection event (NDB_LE_Disconnected ).
(Bug#22773)
SELECT statements with a
BLOB or
TEXT column in the selected
column list and a WHERE condition including a
primary key lookup on a VARCHAR
primary key produced empty result sets.
(Bug#19956) Disk Data:
MEDIUMTEXT columns of Disk Data
tables were stored in memory rather than on disk, even if the
columns were not indexed.
(Bug#25001) Disk Data:
Performing a node restart with a newly dropped Disk Data table
could lead to failure of the node during the restart.
(Bug#24917) Disk Data:
When restoring from backup a cluster containing any Disk Data
tables with hidden primary keys, a node failure resulted which
could lead to a crash of the cluster.
(Bug#24166) Disk Data:
Repeated CREATE , DROP , or
TRUNCATE TABLE in various
combinations with system restarts between these operations could
lead to the eventual failure of a system restart.
(Bug#21948) Disk Data:
Extents that should have been available for re-use following a
DROP TABLE operation were not
actually made available again until after the cluster had
performed a local checkpoint.
(Bug#17605) Cluster API:
Invoking the NdbTransaction::execute() method
using execution type Commit and abort option
AO_IgnoreError could lead to a crash of the
transaction coordinator (DBTC ).
(Bug#25090) Cluster API:
A unique index lookup on a nonexistent tuple could lead to a
data node timeout (error 4012).
(Bug#25059) Cluster API:
When using the NdbTransaction::execute()
method, a very long timeout (greater than 5 minutes) could
result if the last data node being polled was disconnected from
the cluster.
(Bug#24949) Cluster API:
Due to an error in the computation of table fragment arrays,
some transactions were not executed from the correct starting
point.
(Bug#24914)
|
|