[pgpool-general: 7229] Re: Query regarding failover and recovery
ishii at sraoss.co.jp
Sun Aug 30 21:47:28 JST 2020
>> That is an expected behavior. The node previously brought down is left
>> as "down" by pgoool. This is intentional. You need to issue
>> pcp_recovery_node against the node (previous primary node in your
>> case) to make it online again.
> Hi Tatsuo,
> is there a way to have this made automatically?
> I know that this could not be safe in your perspective, but I also could
> agree that if the PostgreSQL process starts again on the old primary it
> means that the problem has been solved, and we want to restore production
> as soon as possible (automatically, if possible).
> Is there some workaround for this?
> Thank you very much!
> if the PostgreSQL process starts again on the old primary it
> means that the problem has been solved,
Not sure. For example, it is possible that the PostgreSQL has started
but as a primary server, not standby server. So just PostgreSQL has
succeeded in starting does not necessary means the PostgreSQL can be
safely attached to the streaming replication cluster.
Or do you mean that the primary server temporally goes down and after
the restarting you want to use the server as the primary again? If so,
the easiest way would be using DISALLOW_TO_FAILOVER flag. See the
manual for more details.
SRA OSS, Inc. Japan
More information about the pgpool-general