0

I installed PgPool-II and nodes on Debian virtual machines. I followed instructions written in the user manual. I enabled streaming replication, failover and online recovery features.

Here is my network configuration :

  • 192.168.1.111 PgPool-II
  • 192.168.1.124 master node (id=0)
  • 192.168.1.126 slave node (id=1)

PgPool-II working well, however I have an issue during online recovery with master node.

When I bring down master node the slave is well promoted (he becomes the new master) due to failover. So far, so good.

However, when and try to restore old master as slave with "pcp_recovery_node" command, this one is correctly reconfigured (due to "basebackup.sh") but it can't connect to new master because "max_wal_senders = 0" on latter.

I tried to change this value on "postgresql.conf" standby file but if I do that my standby doesn't want to start because "wal_level = minimal" and so there is inconsistency.

Versions : - PgPool-II 3.3.3 - PostgreSQL 9.1 - Debian x64 7.6

Thanks in advance for your help.

Antonin
  • 181
  • 2
  • 11
  • 1
    Quoth the docs: wal_level must be set to archive or hot_standby to allow connections from standby servers. So why not just bump wal_level to 'hot_standby'? – Josh Kupershmidt Aug 29 '14 at 21:26
  • @JoshKupershmidt I just tried to bump wal_level to hot_standby (and max_wal_senders to 1) on postgresql.conf standby file as you suggest, and the postgreSQL servers launching well. Thanks a lot ! However this time, my restored old master as slave always seems unable to connect to promoted slave (new master). – Antonin Sep 01 '14 at 07:47

0 Answers0