1
I'm using PostgreSQL 8.4 and am creating a standby server. Since PostgreSQL 8.4 doesn't have built-in replication support, I did the following things try to get read only access to the standby server, and not to transfar full backup everyday.
Set master to transfer WAL files to the standby server in
postgresql.conf
archive_mode = on archive_command = 'scp -C %p standby:/var/lib/pgsql/backups/%f'
create a
recovery.conf
on the standby serverrestore_command = 'mv /var/lib/pgsql/backups/%f %p' recovery_end_command = 'mv /var/lib/pgsql/data/recovery.{done,conf}'
pg_ctl restart -m fast
once a day
Then I realized that PostgreSQL did restore the transfered WAL files at the initial start up, but does not after a restart. I found the next message in the log file.
LOG: selected new timeline ID: 2
It seems that standby server is already on a new timeline, but master server is still sending old timeline's WAL file, which makes standby server not restoring.
So is there a way to make it possible to achieve what I'm trying to do?
Thanks.
sDid you promote the standby at some point? It can't replay the \ – Craig Ringer – 2015-03-26T07:49:38.327
It reads like you'd want to continue restoring into a secondary server that at some point has become a live server? That's not possible because as soon as the end of the recovery, it is allowed to diverge irrevocably from the primary. The new timeline is just a sign indicating that. – Daniel Vérité – 2015-03-27T21:27:36.087