Thread: postmaster service seems to stop without reason

Hello!

Another problem just occurred: We've got a postgres database running
okay, we can access it an everything seems fine - most of the time. It
seems however, that the postmaster stops sometimes and restarts without
our notice - we had a look over the event-log and found several entries
"postmaster stopped" or "postmaster started" - but neither of us
restarted nor rebootet the server, no one stopped or restarted the
service.

This in itself wouldn't be much of a problem if no one would notice. It
seems however, that the postmaster sometimes doesn't delete it's .pid
file, thus refusing to start up again on the next occasion.

I have got a workaround in place which just deletes the .pid before
restarting the service should it fail, but this seems to be to be a
quick and dirty solution: I'd like to know why it stops in the first
place - and why it sometimes doesn't delete it's .pid when it does so.
Is it meant to behave like this?

Any ideas?

Thanks,

    Markus

The most common occasion of not deleting PID files is when a reboot happens
as then a SIGHUP is sent to the postmaster as well as the normal signal.
This causes the postmaster to reread its configuration, which means its busy
and ends up just killed of, therefore not deleteing its .pid file. I believe
that there was a patch that basically deleted its SIGHUP handler. This may
or maynot help.
I've not come across your specific problem in NT, although I've suffered
similar on 9x but never found exactly why (I'd asumed it was just 9x
dropping the ball).
- Stuart

> -----Original Message-----
> From: Markus Wollny [mailto:Markus.Wollny@computec.de]
> Sent: 22 February 2002 13:59
> To: pgsql-cygwin@postgresql.org
> Subject: [CYGWIN] postmaster service seems to stop without reason
>
>
> Hello!
>
> Another problem just occurred: We've got a postgres database running
> okay, we can access it an everything seems fine - most of the time. It
> seems however, that the postmaster stops sometimes and
> restarts without
> our notice - we had a look over the event-log and found
> several entries
> "postmaster stopped" or "postmaster started" - but neither of us
> restarted nor rebootet the server, no one stopped or restarted the
> service.
>
> This in itself wouldn't be much of a problem if no one would
> notice. It
> seems however, that the postmaster sometimes doesn't delete it's .pid
> file, thus refusing to start up again on the next occasion.
>
> I have got a workaround in place which just deletes the .pid before
> restarting the service should it fail, but this seems to be to be a
> quick and dirty solution: I'd like to know why it stops in the first
> place - and why it sometimes doesn't delete it's .pid when it does so.
> Is it meant to behave like this?
>
> Any ideas?
>
> Thanks,
>
>     Markus
>
> ---------------------------(end of
> broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to
> majordomo@postgresql.org
>