In addition to the central configuration file, a cluster may also be controlled through a command-line interface available through the management client ndb_mgm. This is the primary administrative interface to a running cluster.
Commands for the event logs are given in Section 6.3, “Event Reports Generated in NDB Cluster”; commands for creating backups and restoring from them are provided in Section 6.8, “Online Backup of NDB Cluster”.
Using ndb_mgm with MySQL Cluster Manager.
MySQL Cluster Manager handles starting and stopping processes and tracks their
states internally, so it is not necessary to use
ndb_mgm for these tasks for an NDB Cluster
that is under MySQL Cluster Manager control. it is recommended
not to use the ndb_mgm
command-line client that comes with the NDB Cluster distribution
to perform operations that involve starting or stopping nodes.
These include but are not limited to the
SHUTDOWN commands. For more
information, see MySQL Cluster Manager Process Commands.
The management client has the following basic commands. In the
listing that follows,
either a data node ID or the keyword
indicates that the command should be applied to all of the
cluster's data nodes.
Connects to the management server indicated by the connection string. If the client is already connected to this server, the client reconnects.
Creates a new NDB Cluster node group and causes data nodes to join it.
This command is used after adding new data nodes online to an NDB Cluster, and causes them to join a new node group and thus to begin participating fully in the cluster. The command takes as its sole parameter a comma-separated list of node IDs—these are the IDs of the nodes just added and started, and that are to join the new node group. The list must contain no duplicate IDs. The number of nodes in the list must be the same as the number of nodes in each node group that is already part of the cluster (each NDB Cluster node group must have the same number of nodes). In other words, if the NDB Cluster consists of 2 node groups having 2 data nodes each, then the new node group must also have 2 data nodes.
The node group ID of the new node group created by this command is determined automatically, and always the next highest unused node group ID in the cluster; it is not possible to set it manually.
For more information, see Section 6.7, “Adding NDB Cluster Data Nodes Online”.
Drops the NDB Cluster node group with the given
This command can be used to drop a node group from an NDB Cluster.
DROP NODEGROUPtakes as its sole argument the node group ID of the node group to be dropped.
DROP NODEGROUPacts only to remove the data nodes in the effected node group from that node group. It does not stop data nodes, assign them to a different node group, or remove them from the cluster's configuration. A data node that does not belong to a node group is indicated in the output of the management client
no nodegroupin place of the node group ID, like this (indicated using bold text):
id=3 @10.100.2.67 (5.6.51-ndb-7.4.39, no nodegroup)
Prior to NDB 7.0.4, the
SHOWoutput was not updated correctly following
DROP NODEGROUP. (Bug #43413)
DROP NODEGROUPworks only when all data nodes in the node group to be dropped are completely empty of any table data and table definitions. Since there is currently no way using ndb_mgm or the mysql client to remove all data from a specific data node or node group, this means that the command succeeds only in the two following cases:
TRUNCATE TABLEdoes not work for this purpose because this removes only the table data; the data nodes continue to store an
NDBCLUSTERtable's definition until a
DROP TABLEstatement is issued that causes the table metadata to be dropped.
For more information about
DROP NODEGROUP, see Section 6.7, “Adding NDB Cluster Data Nodes Online”.
Enters single user mode, whereby only the MySQL server identified by the node ID
node_idis permitted to access the database.
Exits single user mode, enabling all SQL nodes (that is, all running mysqld processes) to access the database.Note
It is possible to use
EXIT SINGLE USER MODEeven when not in single user mode, although the command has no effect in this case.
Displays information on all available commands.
Terminates the management client.
This command does not affect any nodes connected to the cluster.
Displays a report of type
report-typefor the data node identified by
node_id, or for all data nodes using
Currently, there are three accepted values for
BackupStatusprovides a status report on a cluster backup in progress
MemoryUsagedisplays how much data memory and index memory is being used by each data node as shown in this example:
ndb_mgm> ALL REPORT MEMORY Node 1: Data usage is 5%(177 32K pages of total 3200) Node 1: Index usage is 0%(108 8K pages of total 12832) Node 2: Data usage is 5%(177 32K pages of total 3200) Node 2: Index usage is 0%(108 8K pages of total 12832)
This information is also available from the
EventLogreports events from the event log buffers of one or more data nodes.
report-typeis case-insensitive and “fuzzy”; for
MemoryUsage, you can use
MEMORY(as shown in the prior example),
memory, or even simply
mem). You can abbreviate
BackupStatusin a similar fashion.
Restarts the data node identified by
node_id(or all data nodes).
RESTARTcauses the data node to perform an initial restart; that is, the node's file system is deleted and recreated. The effect is the same as that obtained from stopping the data node process and then starting it again using ndbd
--initialfrom the system shell.Note
Backup files and Disk Data files are not removed when this option is used.
-noption causes the data node process to be restarted, but the data node is not actually brought online until the appropriate
STARTcommand is issued. The effect of this option is the same as that obtained from stopping the data node and then starting it again using ndbd
-nfrom the system shell.
-acauses all current transactions relying on this node to be aborted. No GCP check is done when the node rejoins the cluster.
RESTARTfails if taking the node offline would result in an incomplete cluster. The
-foption forces the node to restart without checking for this. If this option is used and the result is an incomplete cluster, the entire cluster is restarted.
Displays basic information about the cluster and cluster nodes. For all nodes, the output includes the node's ID, type, and
NDBsoftware version. If the node is connected, its IP address is also shown; otherwise the output shows
not connected, accepting connect from, with
any hostused for nodes that are permitted to connect from any address.
In addition, for data nodes, the output includes
startingif the node has not yet started, and shows the node group of which the node is a member. If the data node is acting as the master node, this is indicated with an asterisk (
Consider a cluster whose configuration file includes the information shown here (possible additional settings are omitted for clarity):
[ndbd default] DataMemory= 128G NoOfReplicas= 2 [ndb_mgmd] NodeId=50 HostName=198.51.100.150 [ndbd] NodeId=5 HostName=198.51.100.10 DataDir=/var/lib/mysql-cluster [ndbd] NodeId=6 HostName=198.51.100.20 DataDir=/var/lib/mysql-cluster [ndbd] NodeId=7 HostName=198.51.100.30 DataDir=/var/lib/mysql-cluster [ndbd] NodeId=8 HostName=198.51.100.40 DataDir=/var/lib/mysql-cluster [mysqld] NodeId=100 HostName=198.51.100.100 [api] NodeId=101
After this cluster (including one SQL node) has been started,
SHOWdisplays the following output:
ndb_mgm> SHOW Connected to Management Server at: localhost:1186 Cluster Configuration --------------------- [ndbd(NDB)] 4 node(s) id=5 @198.51.100.10 (5.6.51-ndb-7.4.39, Nodegroup: 0, *) id=6 @198.51.100.20 (5.6.51-ndb-7.4.39, Nodegroup: 0) id=7 @198.51.100.30 (5.6.51-ndb-7.4.39, Nodegroup: 1) id=8 @198.51.100.40 (5.6.51-ndb-7.4.39, Nodegroup: 1) [ndb_mgmd(MGM)] 1 node(s) id=50 @198.51.100.150 (5.6.51-ndb-7.4.39) [mysqld(API)] 2 node(s) id=100 @198.51.100.100 (5.6.51-ndb-7.4.39) id=101 (not connected, accepting connect from any host)
The output from this command also indicates when the cluster is in single user mode (see the description of the
ENTER SINGLE USER MODEcommand, as well as Section 6.6, “NDB Cluster Single User Mode”).
This command does not shut down any SQL nodes or API nodes that are connected to the cluster.
Brings online the data node identified by
node_id(or all data nodes).
ALL STARTworks on all data nodes only, and does not affect management nodes.Important
To use this command to bring a data node online, the data node must have been started using
Displays status information for the data node identified by
node_id(or for all data nodes).
Possible node status values include
The output from this command also indicates when the cluster is in single user mode (status
SINGLE USER MODE).
Stops the data or management node identified by
ALL STOPworks to stop all data nodes only, and does not affect management nodes.
-aoption causes the node to be stopped immediately, without waiting for the completion of any pending transactions.
STOPfails if the result would cause an incomplete cluster. The
-foption forces the node to shut down without checking for this. If this option is used and the result is an incomplete cluster, the cluster immediately shuts down.Warning
Use of the
-aoption also disables the safety check otherwise performed when
STOPis invoked to insure that stopping the node does not cause an incomplete cluster. In other words, you should exercise extreme care when using the
-aoption with the
STOPcommand, due to the fact that this option makes it possible for the cluster to undergo a forced shutdown because it no longer has a complete copy of all data stored in
Additional commands. A number of other commands available in the ndb_mgm client are described elsewhere, as shown in the following list:
START BACKUPis used to perform an online backup in the ndb_mgm client; the
ABORT BACKUPcommand is used to cancel a backup already in progress. For more information, see Section 6.8, “Online Backup of NDB Cluster”.
CLUSTERLOGcommand is used to perform various logging functions. See Section 6.3, “Event Reports Generated in NDB Cluster”, for more information and examples.
For testing and diagnostics work, the client also supports a
DUMPcommand which can be used to execute internal commands on the cluster. It should never be used in a production setting unless directed to do so by MySQL Support. For more information, see NDB Cluster Management Client DUMP Commands.