From: "Oscar Serrano" <oserra@fondos.net>
To: <pgsql-interfaces@postgresql.org>
Subject: RE: Backend dies for DBI 
Date: Thu, 8 Mar 2001 11:15:30 +0100
Message-ID: <LLEKJGBGFGPBMIGHIFFOAEJDCEAA.oserra@fondos.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
In-Reply-To: <24449.984016287@sss.pgh.pa.us>
Importance: Normal


> -----Mensaje original-----
> De: pgsql-interfaces-owner@postgresql.org
> [mailto:pgsql-interfaces-owner@postgresql.org]En nombre de Tom Lane
> Enviado el: jueves, 08 de marzo de 2001 2:51
> Para: Bjorn Halvor Solberg
> CC: pgsql-interfaces@postgresql.org
> Asunto: Re: [INTERFACES] Backend dies for DBI
>
>
> > What confuses me is that it ran fine for a couple of months.
> Then, after
> > the power outage, it crashes every few hours, resulting in having to
> > restart the web server to regain the DB handle.
>
> Hmm.  Possibly some corrupted data in the database now?
>
> 			regards, tom lane
>

Recently I had a problem similar to this.
I don't know why, once or twice a day, some backend died abnormaly so
killing all other backends. The problem was some index got corrupted. Using
the command REINDEX TABLE table_name FORCE, it fixed the problem with the
index. But the problem started again.
Finally I've found that the problem was in NFS. My databases where in
another computer (different from the postgres binaries) mounted by NFS. Now
I've put postmaster and databases in the same hard disk and now it works.

BTW, I have a conexion to the postmaster on every cgi (mod_perl) because I
thought that it was not possible to open a conexión from within the
httpd.conf and use it for every cgi.

> ---------------------------(end of broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
>