Re: 9.1.1 crash - Mailing list pgsql-general

From Albe Laurenz
Subject Re: 9.1.1 crash
Date
Msg-id D960CB61B694CF459DCFB4B0128514C207779418@exadv11.host.magwien.gv.at
Whole thread Raw
In response to 9.1.1 crash  (Mike Blackwell <mike.blackwell@rrd.com>)
List pgsql-general
Mike Blackwell wrote:
> The following are the relevant log entries from a recent crash of v9.1.1 running on an older RHEL
> Linux box.  This is the first crash we've experienced in a lot of years of running Pg.  Any assistance
> in how to determine what might have caused this is welcome.
> 
> 2012-02-10 13:55:59 CST [15949]: [37-1] @ LOG:  00000: server process (PID 32670) was terminated by
> signal 11: Segmentation fault
[...]

It is difficult to find out anything after the crash if the problem
cannot be reproduced.

If you happen to have changed the core file ulimit setting away from the
default zero, you should have a core file in the data directory which
can be used to create a backtrace which shows you where the server
crashed. And even that only really helps with a debug build.

Other than that, you could make sure that hard disk and memory have
no problem (you write that it is an older box). You can try to find
out what the server was doing at the time and if you can reproduce it.

Crashes are also often caused by nonstandard C funxtions that have
been loaded into the database.

Yours,
Laurenz Albe

pgsql-general by date:

Previous
From: Jasen Betts
Date:
Subject: Re: URGENT: temporary table not recognized?
Next
From: Wim Bertels
Date:
Subject: Re: psql latex and newlines