On Fri, 8 Nov 2002, Nick Fankhauser wrote:
> AM-
>
> The band-aid patch is to increase the number of connections available. Check
> out this link to the idocs:
>
> http://www.postgresql.org/idocs/index.php?runtime-config.html
>
> Maybe that will keep things running for you while you go over the code to
> learn why you're using more connections now. If you aren't using pooled
> connections, maybe you just have more users on the web.
But is there really that number of backends/connections present?
Could be a DoS on the web server but if it happens immediately on restart of
everything it must be an ongoing and pretty obvious attack.
I see that error message in storage/ipc/sinval.c (7.2.3) so it could be some
sort kernel limit on shared memory connections or something like that. I
haven't looked further than a grep so I could be completely wrong in that
suggestion.
>
> > -----Original Message-----
> > From: pgsql-admin-owner@postgresql.org
> > [mailto:pgsql-admin-owner@postgresql.org]On Behalf Of A.M.
> > Sent: Friday, November 08, 2002 12:31 PM
> > To: pgsql-admin@postgresql.org
> > Subject: [ADMIN] FATAL 1: Sorry, too many clients already
> >
> >
> > OMG, I'm about to rip my hair out. I've been writing a webapp using
> > DBD::Pg and it has been accessing postgresql for weeks without
> > problems. Suddenly this morning with no changes to the setup, I get:
> > FATAL 1: Sorry, too many clients already
> > whenever I try to use the webapp (running under user "apache"). This
> > message appears when I have psql open and in the httpd error_log.
> > However, if I connect using psql via the domain socket or tcp socket, I
> > have no problems! I restarted postgres, apache, and the entire server
> > and the shit won't go away. What am I doing wrong?!
> > Setup:
> > RedHat 7.1
> > pgsql 7.2.2
> > Nothing other than the db and webapp are running on the server. What
> > should I do?
> > ><><><><><><><><><
> > AgentM
> > agentm@cmu.edu
> >
--
Nigel J. Andrews