Thanks Michael, So everytime after promoting Slave to become master (either manually or automatic), just stop Slave2 and run pg_rewind on slave2 against the new maser (old slave1). And when old master server is available again, use pg_rewind on that serve as well against new master to return to original configuration. -----Original Message----- From: Michael Paquier [mailto:michael@xxxxxxxxxxx] Sent: Tuesday, 6 March 2018 3:54 PM To: Dylan Luong <Dylan.Luong@xxxxxxxxxxxx> Cc: pgsql-generallists.postgresql.org <pgsql-general@xxxxxxxxxxxxxxxxxxxx> Subject: Re: Resync second slave to new master On Tue, Mar 06, 2018 at 04:45:10AM +0000, Dylan Luong wrote: > After a failover (promote) to the Slave1, is it easily resync the > Slave2 to the new master (old slave1)? Do we need to do full rebuild > of the Slave2 from new master everytime we failover to Slave1 from > Master? Can we use pg_rewind on Slave2 to resyn it with new master > (old slave1)? After promoting slave 1, it could be possible that some records have slipped to slave 2 from the primary. In this case, a rewind would be recommended. You should be careful that your slave 2 has not received WAL to a position newer than where WAL has forked because of the promotion. If that happened, then a rewind would be necessary before replugging slave 2 to the newly-promoted server. Be very careful with your failover flow here. pg_rewind also would not run if it finds that the target server does not need a rewind, so you could stop the slave 2, and run pg_rewind unconditionally to keep things simple. -- Michael
