From: | "scott(dot)marlowe" <scott(dot)marlowe(at)ihs(dot)com> |
---|---|
To: | Garo Hussenjian <garo(at)xapnet(dot)com> |
Cc: | Postgresql General <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: Idle transaction causing problems. |
Date: | 2003-02-18 23:24:24 |
Message-ID: | Pine.LNX.4.33.0302181621350.12007-100000@css120.ihs.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Tue, 18 Feb 2003, Garo Hussenjian wrote:
> I have a server (pgsql 7.1.2) that periodically racks up a bunch of UPDATE
> waiting processes causing my php application to hang... I used gdb to get
> the debug_query_string on one of the UPDATE waiting processes and found it
> to be a very simple query on our session table... Not a server-breaker!
>
> The culprit seemed to be another process with status 'transaction idle' but
> the gdb debug_query_string was null (pointed to 0x0)... When I killed the
> transaction idle process the UPDATE waiting processes cleared out
> immediately and we were up and running again...
>
> Is there a way (w/ gdb or other) to determine the source of the idle
> transaction blocking traffic?
Not helping you troubleshoot the exact situation, just useful info on PHP
/ Postgresql...
Fyi, older versions of php had a bug that when they put out a notice
condition, it could cause apache children to crash and burn. One of the
most common notice messages is about supplying a missing from clause or
some such.
This was fixed around 4.2 or 4.3 I believe. To find the culprit, try
running your various queries from a psql interface and look for the
notice. Any query generating a notice can crash the backend, but only
randomly, not every time.
I had a script that was doing something similar, leaving ports open.
From | Date | Subject | |
---|---|---|---|
Next Message | Justin Clift | 2003-02-18 23:37:45 | Re: Configuring postmaster to use all available memory |
Previous Message | Dima Tkach | 2003-02-18 22:58:06 | Re: postgres error reporting |