On Wednesday 15 April 2009 19:21:03 you wrote:
> Janning Vygen <vygen@kicktipp.de> writes:
> > Now i see that the errors occur _exactly_ every 4000 seconds (1 hour, 6
> > minutes and 40 seconds). I have no clue as i only have one cronjob at
> > night concerning postgresql. I have no autovacuum running (only manual at
> > night). my application cronjobs are only running at night. i have a few
> > threads but no thread has a sleep time of 4000 millis, besides the fact
> > that they all work fine. Maybe i have to check my c3p0 pool. however, i
> > know i have to search for my own, as you can't look into my app. But
> > maybe someone has a hint, that something inside or outside postgresql is
> > usually running every 4000 seconds?
>
> There's nothing inside of postgres that would operate on such a
> schedule. Furthermore, unless I'm all wet about where the error is
> coming from, this *must* be happening during receipt of a command from
> an external client.
You were right.
> Perhaps turning on log_connections would give you a clue about what it
> is.
thanks. Turning log_connections on helped me a lot! I found a buggy query from
my postfix mailserver which retried the query every 4000 seconds.
thanks again for this excellent database and excellent support.
kind regards
Janning
> Also set log_line_prefix to make sure you can match up the error
> report with the previous connection log entry (adding the PID is usually
> enough).
>
> regards, tom lane