







|
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MySQL 5.1 tree are no longer maintained; these users should use the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 | |