During fail back, if prepare_failback was executed successfully on the secondary companion but the primary companion does not start, perform the following to roll back and then reissue the prepare_failback command:
Check the primary companion’s system event log to find the reason the server does not start, and correct the problems.
Check that the MSCS group that contains the resource for the primary server is located on the secondary node. If not, it does use Move Group to move it.
Log in to the secondary companion and issue:
dbcc ha_admin ("", "rollback_failback")
dbcc ha_admin ("", "rollback_failover")
Verify that the secondary companion is in normal companion mode.
Check that the MSCS resource for the primary server is online. If it is not, manually bring the resource online using the Cluster Administrator.
