Once a failover has occurred, Data Mover does not automatically failback to the old master daemon if it is restarted. To fail back to the old-restored master after a failover has occurred, do the following:
-
Log on to the old master daemon server that needs to be restored as master and set it as Slave using the command dmcluster setslave.
This starts up the sync service associated with this daemon in slave mode. Once the sync service is started in slave mode, it will get all the updates that occurred on the new master repository while the old master was down and catch up with the new master.
- Wait for a few minutes to allows the old master to catch up with the new master, and then check the dmSync.log on the slave (old master) to determine when the old master has caught up with the new master.
- When the old master repository is in sync with the new master repository, set the old master back as the new master by running the dmcluster setmaster command on the old master.
-
Run the dmcluster status command to verify the old master has been restored successfully.
Trying to start the daemon on the old-restored master using the dmdaemon service script (/etc/init.d/dmdaemon start) will cause the monitoring service to assume that the daemon is incorrectly started on the old master and shut it down.