[pgpool-general: 8489] Re: disable load balancing to a master/slave cluster with streaming replication

[Giuseppe Sacco]

Il giorno mar, 08/11/2022 alle 08.24 +0100, Giuseppe Sacco ha scritto:
> [...]
> > > How may I collect information about the way pgpool select the backend?
> > > Is
> > > there any specific string I should look for in the log when running with
> > > option -d? Does debug output include valuable information?
> > 
> > If you enable "log_per_node_statement = on" to output the queries, it may
> > help.
> 
> I am going to try. Thank you.

I checked the log and found a problem about the status check on the master
backend. Pgpool status for that backend was down, so this explain why all
queries were sent to the slave.

I stopped pgpool, I manually updated /var/log/postgresql/pgpool_status, then I
started pgpool again and now it works. (Should I have used pcp_attach_node
instead?)

Is there any trigger to be notified about a backend being marked down by
pgpool?

Thank you,
Giuseppe