From: | Chris Curvey <chris(at)chriscurvey(dot)com> |
---|---|
To: | Jeff Janes <jeff(dot)janes(at)gmail(dot)com> |
Cc: | Chris Curvey <ccurvey(at)zuckergoldberg(dot)com>, "pgsql-general(at)postgresql(dot)org" <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: help getting a backtrace from 9.2 on Ubuntu 13.04? |
Date: | 2013-09-16 00:09:11 |
Message-ID: | CADfwSsCuWG3rCe_3FmH8oZ=vZ93ttxGj-MHc27T482gQLdueAA@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Sun, Sep 15, 2013 at 7:49 PM, Jeff Janes <jeff(dot)janes(at)gmail(dot)com> wrote:
> On Tue, Sep 10, 2013 at 10:37 AM, Chris Curvey <ccurvey(at)zuckergoldberg(dot)com
> > wrote:
>
>>
>>
>>
>>
>> Great thought. Looking through the logs, it appears that all my failures
>> are on a CREATE INDEX. Usually on my biggest table, but often on another
>> table.
>>
>>
>>
>> 2013-09-10 10:09:46 EDT ERROR: canceling autovacuum task
>>
>> 2013-09-10 10:09:46 EDT CONTEXT: automatic analyze of table
>> "certified_mail_ccc2.public.cm_status_history"
>>
>> 2013-09-10 10:15:13 EDT LOG: server process (PID 14386) was terminated
>> by signal 11: Segmentation fault
>>
>> 2013-09-10 10:15:13 EDT DETAIL: Failed process was running: CREATE INDEX
>> cm_envelope_tracking_number ON cm_envelope USING btree (tracking_number);
>>
>>
>>
>>
>>
>>
>>
>> 2013-09-10 10:15:13 EDT LOG: terminating any other active server
>> processes
>>
>> 2013-09-10 10:15:13 EDT WARNING: terminating connection because of crash
>> of another server process
>>
>> 2013-09-10 10:15:13 EDT DETAIL: The postmaster has commanded this server
>> process to roll back the current transaction and exit, because another
>> server process exited abnormally and possibly corrupted shared memory.
>>
>>
>>
>> I cannot square this with the fact that when I echo the commands, the
>> last echoed command is about setting privileges.
>>
>
> A backend is crashing, and taking down the entire PostgreSQL system. The
> commands you see being echoed are from a different process from the one
> that triggered the crash, so it is just an innocent bystander which has no
> useful information. Are you using parallel restore? (If not, why is there
> someone indexing your biggest table during the restore?)
>
I'm the only person doing anything, and the only thing going on is the
restore. And I'm not using parallel restore (I thought that might be part
of the issue.)
>
> You will want to get the backtrace of the coredump generated by the
> crashed backend, not of the running process. Have you tried taking a bt
> with gdb? You said you couldn't find the symbols, but have you tried it
> anyway? On CentOS and openSuse I often get warnings about some symbols not
> being found, but all the symbols I actually need to interpret the backtrace
> end up being there.
>
I can try, but the instructions said that installing the -dev package by
itself was not sufficient, so I stopped at that point. But as they say in
the lottery ads, "Hey, you never know!"
> Cheers,
>
> Jeff
>
Many thanks!
From | Date | Subject | |
---|---|---|---|
Next Message | Jeff Janes | 2013-09-16 00:30:22 | Re: Segmentation fault: pg_upgrade 9.1 to 9.3: pg_dump: row number 0 is out of range 0..-1 |
Previous Message | Robert Nix | 2013-09-16 00:06:44 | Re: Segmentation fault: pg_upgrade 9.1 to 9.3: pg_dump: row number 0 is out of range 0..-1 |