Are you using extended protocol via JDBC or some such? If so, following might be related:
http://lists.pgfoundry.org/pipermail/pgpool-committers/2009-June/000910.html -- Tatsuo Ishii SRA OSS, Inc. Japan > I have an install of pgpool-2.1 on FreeBSD 6.0 (AMDx64) that had been > happily running for 73 days+ at around 250k unique connections per day > coming from psycopg2 on the local machine. pgpool is set up just to > pool connections to a remote machine -- no HA, no replication, etc. > All connections are made under two different usernames, no more. > > Suddenly, all attempts at connection via psycopg2 started yielding: > > OperationalError: server closed the connection unexpectedly > This probably means the server terminated abnormally > before or while processing the request. > > I was unable to connect via psql, either, through pgpool. I could > connect normally to PostgreSQL on the remote machine. > > pgpool was still running, there was nothing in any logs about any > problems. I didn't have connection logging turned on, but no error > conditions were raised, either, but pgpool stopped responding to new > connection requests. > > I issued a `pgpool stop; pgpool` command and things immediately went > back to normal. > > I'm in the process of testing/upgrading to the more recent version of > pgpool, but I did not see any items in the change logs/NEWS indicating > anything like this was mentioned or resolved. > > Any ideas on how I can prevent this? I'd rather not bandaid over this > with a cron to run `pgpool stop; pgpool`. > _______________________________________________ > Pgpool-general mailing list > [email protected] > http://pgfoundry.org/mailman/listinfo/pgpool-general _______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
