Обсуждение: sig 6 on postgres server

Hi Folks:
Recently I had a postgres instance shut down & restart, and I'm trying
to understand why. I'm running 8.3.5 on solaris 10, with the data
directory on a netapp. The postgres log shows an error associated with
some of the critical application tables not existing, then it was unable
to write to the log, then a sigabrt. Not long after postgres recovered.
There were no errors in the application (other than yelling about not
having a database), no syslog errors on the zone, no errors on the
netapp, no known break in connectivity. I ran a vacuum analyze based on
one posting I saw on this list, and things seem fine.

My questions are
- any idea what happened? Are there other things I should check?
- what should I do to ensure there is no data corruption?

Related pg log entries...
cpas 2009-03-02 14:56:32 PSTERROR:  relation "exp.domainproperty" does
not exist
...
cpas 2009-03-02 14:58:34 PSTPANIC:  could not write to log file 93,
segment 240 at offset 9560064, length 8192: Interrupted system call
@- 2009-03-02 14:58:37 PSTLOG:  server process (PID 19257) was
terminated by signal 6
@- 2009-03-02 14:58:37 PSTLOG:  terminating any other active server
processes
cpas 2009-03-02 14:58:37 PSTWARNING:  terminating connection because of
crash of another server process
Thanks for the help!
Mija

CONFIDENTIALITY NOTICE: This e-mail message, including any attachments,
is for the sole use of the intended recipient(s) and may contain
confidential, proprietary, and/or privileged information protected by
law. If you are not the intended recipient, you may not use, copy, or
distribute this e-mail message or its attachments. If you believe you
have received this e-mail message in error, please contact the sender by
reply e-mail and destroy all copies of the original message.

"Lee, Mija" <mija@scharp.org> writes:
> Recently I had a postgres instance shut down & restart, and I'm trying
> to understand why. I'm running 8.3.5 on solaris 10, with the data
> directory on a netapp. The postgres log shows an error associated with
> some of the critical application tables not existing, then it was unable
> to write to the log, then a sigabrt.

> cpas 2009-03-02 14:58:34 PSTPANIC:  could not write to log file 93,
> segment 240 at offset 9560064, length 8192: Interrupted system call

EINTR?  Hmm, I wonder if you have the data directory on a "soft" NFS
mount?  We tend not to trust NFS too much at all around here, but if
you have to use it, hard-mount is more reliable than soft.

            regards, tom lane

Thanks for the response, Tom. We do have hard mounts, and unfortunately,
we are dependent on nfs.

-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Monday, March 02, 2009 6:47 PM
To: Lee, Mija
Cc: pgsql-novice@postgresql.org
Subject: Re: [NOVICE] sig 6 on postgres server

"Lee, Mija" <mija@scharp.org> writes:
> Recently I had a postgres instance shut down & restart, and I'm trying
> to understand why. I'm running 8.3.5 on solaris 10, with the data
> directory on a netapp. The postgres log shows an error associated with
> some of the critical application tables not existing, then it was
unable
> to write to the log, then a sigabrt.

> cpas 2009-03-02 14:58:34 PSTPANIC:  could not write to log file 93,
> segment 240 at offset 9560064, length 8192: Interrupted system call

EINTR?  Hmm, I wonder if you have the data directory on a "soft" NFS
mount?  We tend not to trust NFS too much at all around here, but if
you have to use it, hard-mount is more reliable than soft.

            regards, tom lane