DRBD

Posted Nov 21, 2010 0:47 UTC (Sun) by butlerm (subscriber, #13312)
In reply to: DRBD by kjp
Parent article: Comparing MySQL and Postgres 9.0 Replication (TheServerSide)

What I want to know is if DRBD and/or PostgreSQL has "catch up" capability when you want to switch back. After a false alarm, or temporary outage for example. Or do you have to shutdown the secondary, take a backup, and restore it on the primary?

DRBD

Posted Nov 21, 2010 10:41 UTC (Sun) by dlang (guest, #313) [Link]

since the postgres built-in replication is shipping the WAL log over, if you get just a bit behind you should be able to catch up.

however the server doesn't keep the WAL log around forever, so I would expect that if you get so far behind that the master no longer has a copy of things you have to do a backup.

note that this is speculation, I'm not an expert here

DRBD

Posted Nov 21, 2010 18:39 UTC (Sun) by ewan (guest, #5533) [Link]

I believe that as far as DRBD goes you can bring up the failed system as the slave, and just have it sync from the master in the normal manner until the next failover. Or, in short, 'Yes'.

I think.