[Pgpool-general] Re: Bad degeneration when shutting down secondary backend

[Tatsuo Ishii]

> > When PostgreSQL goes down, any transaction receives an error status
> > "FATAL: terminating connection due to administrator command" from
> > it. That's why you see:
> > 
> > > pool_process_query: kind does not match between backends 
> > master(C) secondary(E)
> > 
> > Here "secondary(E)" means the secondary backend returns the "FATAL:
> > terminating..." error.
> > --
> > Tatsuo Ishii
> > 
> Ok, that's exactly what is happening. After testing deeper, we found that
> stopping the "secondary" postgresql backend rises the problem. Reboot and
> shutdown, invoke the stop, that's why the problem happened there.
> 
> I think pgpool should understand the "FATAL: terminating connection due to
> administrator command" as a backend going down, and shut that backend down
> internally. What do you think?

Actuall I tried this long time ago. Problem was sometimes pgpool could
not read the entire message probably because postmaster closes the
connection. 

> In the while, as I said in a previous mail, a workaround is to close the
> port TCP:5432 using iptables in the database machine previous to shut down
> postgresql (or the entire machine).
--
Tatsuo Ishii