We are doing a DRP test on a large database, which is hosted on an AlwaysOn availablilty group with 3 replicas (2 sync, 1 async). The plan is to failover to the asynch replica, make some tests, then discard this database and revert back to the synchronous replicas and then resynch back to the asynch replica. Because this test is a little different than an actual DR scenario, I would like to get a confirmation of the steps to failover and failback. I'd rather not have to restore the entire AG group, keeping the original 2 synchronous replicas.
Let's call these replicas P1 (primary replica), P2 (secondary replica, synchronous), DR (secondary replica, asynchronous).
Stop/disable SQL service for P2
Stop/disable SQL service for P1
Force failover on DR (ALTER AVAILABILITY GROUP agxxxx FORCE_FAILOVER_ALLOW_DATA_LOSS)
-- MAKE TESTS --
Remove DR from availabililty group
Delete DR replica database
Start/enable SQL service for P1
Start/enable SQL service for P2
Force failover on P1 (ALTER AVAILABILITY GROUP agxxxx FORCE_FAILOVER_ALLOW_DATA_LOSS)
Add DR back to availabililty group, resynchronize database
Will it work like this? I'm a bit unclear at what to expect when we failback. If the asynch replica is removed first (to prevent test data from going into prod), will the synchronous replicas resync correctly and come online when I force failover to P1?