[Pgpool-general] pgpool-II does not re-establish connection to database

[Jaume Sabater]

On Mon, Jan 19, 2009 at 10:46 PM, Aleksey Tsalolikhin
<atsaloli.tech at gmail.com> wrote:

> pgpool-II fails to reconnect to the database after the database goes
> down and then comes back up.
>
> Instead, pgpool-II clients get message like,
>
>        Server closed the connection unexpectedly
>
>        This probably means server terminated abnormally before or
> while processing the request

This happens when the master server is the one failing. If it were the
slave, it would failover the node and execute the query in the master
node. Unfortunately, there is a bug in version 2.1 that made this
happen, but in version 2.2 when the master node fails, the same
behaviour will occur. This behaviour is intended, as explained in this
thread:

http://lists.pgfoundry.org/pipermail/pgpool-general/2008-July/001119.html

> The only way I can get pgppool-II to reconnect is to manually restart
> the pgpool-II instance.
>
> Is it possible to configure pgpool-II to attempt to reconnect to the
> database server after the server
> closes the connection unexpectedly?

Once a node goes down and if pgpool-II detects it, you really don't
want it to just assume no changes happened and reattaching the node
just like that. You want to do an online recovery.

-- 
Jaume Sabater
http://linuxsilo.net/

"Ubi sapientas ibi libertas"