I have two PgSQL servers, both running 9.4. Slave (B) async replicates master (A).
If at some point, A fails and I failover to B, the slave-turned-master creates a new timeline and can begin serving requests. However, if I'm able to restore service on A quickly, I'd like to bring it back into service as master (knowing that any transactions against B would be lost). If I'm okay with those losses (or if potentially I can bring A back before any additional transactions are committed), B still can't resume replication because it's forked the timeline. What I'm hoping for is a way to undo that timeline fork and allow B to resume replicating A without a full (and lengthy) basebackup.
I know there's pg_rewind in 9.5+ but I'm not there yet. Is what I'm trying to do possible? What are the steps for manually reverting to a slightly early point in time on B purely so I can resume replication?
thanks!