3.2.1 Starting and Stopping the Agent on Linux

To start the MySQL Cluster Manager agent on a given host running a Linux or similar operating system, you should run mcmd, found in the bin directory within the manager installation directory on that host. Typical options used with mcmd are shown here:

mcmd [--defaults-file | --bootstrap] [--log-file] [--log-level]

See Section 3.1, “mcmd, the MySQL Cluster Manager Agent”, for information about additional options that can be used when invoking mcmd from the command line, or in a configuration file.

mcmd normally runs in the foreground. If you wish, you can use your platform's usual mechanism for backgrounding a process. On a Linux system, you can do this by appending an ampersand character (&), like this (not including any options that might be required):

$> ./bin/mcmd &

By default, the agent assumes that the agent configuration file is etc/mcmd.ini, in the MySQL Cluster Manager installation directory. You can tell the agent to use a different configuration file by passing the path to this file to the --defaults-file option, as shown here:

$> ./bin/mcmd --defaults-file=/home/mcm/mcm-agent.conf

The --bootstrap option causes the agent to start with default configuration values, create a default one-machine cluster named mycluster, and start it. This option works only if no cluster has yet created, and is mutually exclusive with the --defaults-file option. Currently, any data stored in the default cluster mycluster is not preserved between cluster restarts; this is a known issue, which we may address in a future release of MySQL Cluster Manager.

The use of the --bootstrap option with mcmd is shown here on a system having the host name torsk, where MySQL Cluster Manager has been installed to /home/jon/mcm:

$> ./mcmd --bootstrap
MySQL Cluster Manager 1.4.8 started
Connect to MySQL Cluster Manager by running "/home/jon/mcm/bin/mcm" -a torsk:1862
Configuring default cluster 'mycluster'...
Starting default cluster 'mycluster'...
Cluster 'mycluster' started successfully
        ndb_mgmd        torsk:1186
        ndbd            torsk
        ndbd            torsk
        mysqld          torsk:3306
        mysqld          torsk:3307
        ndbapi          *
Connect to the database by running "/home/jon/mcm/cluster/bin/mysql" -h torsk -P 3306 -u root

You can then connect to the agent using the mcm client (see Section 3.3, “Starting the MySQL Cluster Manager Client”), and to either of the MySQL Servers running on ports 3306 and 3307 using mysql or another MySQL client application.

The --log-file option allows you to override the default location for the agent log file (normally mcmd.log, in the MySQL Cluster Manager installation directory).

You can use --log-level option to override the log-level set in the agent configuration file.

See Section 3.1, “mcmd, the MySQL Cluster Manager Agent”, for more information about options that can be used with mcmd.

The MySQL Cluster Manager agent must be started on each host in the MySQL NDB Cluster to be managed.

To stop one or more instances of the MySQL Cluster Manager agent, use the stop agents command in the MySQL Cluster Manager client. If the client is unavailable, you can stop each agent process using the system's standard method for doing so, such as ^C or kill.

You can also set the agent up as a daemon or service on Linux and other Unix-like systems. (See Section 2.3.1, “Installing MySQL Cluster Manager on Unix Platforms”.) If you also want data node failed processes from a running MySQL NDB Cluster to be started when the agent fails and restarts in such cases, you must make sure that StopOnError is set to 0 on each data node (and not to 1, the default).