5.10.6. Resetting an Entire Dataservice from Filesystem Snapshots

To restore an entire dataservice from filesystem snapshots, the steps below should be followed. The same snapshot should be used on each host so that data on each host is the same. The following steps should be followed:

  1. Set the dataservice into the MAINTENANCE policy mode:

    [LOGICAL:EXPERT] /alpha > set policy maintenance
  2. The following steps must be completed on each server before completing the next step:

    1. Stop the Tungsten Clustering services:

      shell> stopall
    2. Stop MySQL:

      shell> sudo /etc/init.d/mysql stop
    3. Replace the MySQL data files with the filesystem or snapshot data.

    4. Delete the THL files for each of the services that need to be reset:

      shell> rm /opt/continuent/thl/alpha/*
    5. Start MySQL to perform maintenance on the Tungsten schemas:

      shell> sudo /etc/init.d/mysql start
    6. Delete any Tungsten service schemas:

      mysql> DROP DATABASE tungsten_alpha;

    Once these steps have been executed on all the servers in the cluster, the services can be restarted.

  3. On the current master, start the Tungsten Clustering services:

    shell> startall

    Now start the services using the same command on each of the remaining servers.