On Tue, Jan 24, 2012 at 12:43 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Phil Sorber <phil@omniti.com> writes:
>> I've attached a backtrace.
>
> How about a test case? I have no faith in the proposed patch without
> some evidence of what it's supposed to fix.
We are having trouble coming up with a test case that can reliably
reproduce this. I has happened 5 times since Friday. There was another
one this morning and we have more logging turned on now and I have
attached more information. Is there anything else we can do to help
debug it? This is a production machine and we may be forced into
applying the patch we came up with, but we wanted to have as good of a
solution as possible.
>
> regards, tom lane
I have a new backtrace which is almost identical to the old one. Only
memory addresses differ. I am attaching it anyway. I am also attaching
the pertinent logs with sensitive information changed. The log lines
are just the entries for the PID that segfaulted from right before it
happened. The PID was long lived but nothing was going on for minutes
of time before the segfault. I am also attaching the source of the
function that was called right before the segfault. Again with
sensitive information changed.
Thanks.