3.8 Restoring a MySQL Cluster Manager Agent with Data from Other Agents

Sometimes, an mcmd agent can fail to restart after a failure because its configuration store has been corrupted (for example, by an improper shutdown of the host). If there is at least one other mcmd agent that is still functioning properly on another host of the cluster, you can restore the failed agent by the following steps:

For MySQL Cluster Manager 1.4.6 and earlier:

• Make sure the mcmd agent has really been stopped.

• Go to the agent repository (the agent's mcm_data folder).

• Wipe the contents of the rep folder.

• Delete the metadata files high_water_mark and repchksum.

• Delete the manager.lck file.

• Restart the agent.

For MySQL Cluster Manager 1.4.7 and later:

• Make sure the mcmd agent has really been stopped.

• Restart mcmd with the --initial option, which wipes the contents of the rep folder after backing them up and then starts the agent.

The agent then recovers the configuration store from other agents on the other hosts.

However, if all the mcmd agents for the cluster are malfunctioning, you will have to do one of the following:

• Restore one of the agents first using an agent backup (see Section 3.7, “Backing Up and Restoring MySQL Cluster Manager Agents” for details), and then restore the other agents with it .

• Recreate the whole cluster and restore it using a cluster backup (see Section 3.6, “MySQL NDB Cluster Backup and Restore Using MySQL Cluster Manager” for details).