> On Aug 4, 2018, at 13:50, Michael Paquier <michael@paquier.xyz> wrote:
>
> Hm?
The specific situation is if pg_rewind is attached to the target before the forced post-recovery checkpoint completes,
thetarget can be corrupted:
https://www.postgresql.org/message-id/ECE3B665-E9DD-43FF-B6A6-734E7435221D@thebuild.com
--
-- Christophe Pettus
xof@thebuild.com