Hello. We have multiple SQL Always On Availability Groups that will have their primary nodes/replicas migrated with Zerto. These AGs are in asynchronous commit mode with manual failover (secondaries are for DR and some data loss is acceptable). We’ve migrated non-AG sql vms with no problems, but our concern is the potential rollback to an old primary node/vm if post-migration testing fails…the secondary would contain data from the testing that the old primary does not have and replication would be out of sync.
Seems like we have three options. 1) reverse protect from the new primary to the old primary after migration, so the old primary would be up to date with the secondary’s testing data, 2) power off the secondary before migration and leave it off during testing, and if testing is successful, power it on and it will catch up with the new primary, and if testing failed it will still be in sync with the old primary, 3) remove all dbs from the AGs and add them back after migration (successful or failed).
Option 2 sounds like the path of least resistance with the lowest potential for the nodes to get out of sync. Option 3 would be a pita since the dbs are pretty large and there are many of them.
Gary S.
