From: | Heikki Linnakangas <hlinnakangas(at)vmware(dot)com> |
---|---|
To: | Pierre Slania <pierre(dot)slania(at)lampiris(dot)be>, "pgsql-bugs(at)postgresql(dot)org" <pgsql-bugs(at)postgresql(dot)org> |
Subject: | Re: Out of memory: postgres killed after changing "checkoint_completion_target" and reloading postgres configuration file |
Date: | 2014-12-09 14:07:27 |
Message-ID: | 5487021F.4070900@vmware.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-bugs |
On 12/08/2014 12:47 PM, Pierre Slania wrote:
> Problem :
> ------------
> Dec 4 10:42:41 "myserver" kernel: [7604991.147522] Out of memory: Kill process 2904 (postgres) score 217 or sacrifice child
>
> How to cause the error:
> -----------------------------
> I changed the parameter "checkpoint_completion_target" from 0.5 to 0.9 in the postgresql.conf file unless the production is up and used by multiple users. After that I made a postgres reload the 4th dec 2014 at 10:20AM
>
> (Note that same problem has occurred several weeks ago when I changed "checkpoint_segment" from 3 to 10 in the postgresql.conf file after reloading postgres)
>
> Application that are running
> -----------------------------------
> AS JBoss 7.1.1.Final with a connection pool
>
> Postgres version running in production
> ------------------------------------------------
> PostgreSQL 9.1.8 on x86_64-unknown-linux-gnu, compiled by gcc-4.4.real (Debian 4.4.5-8) 4.4.5, 64-bit
>
> OS Server production
> --------------------------
> Linux "myserver" 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux
>
> The following memories value was done today:
>
> postgres@"myserver":/var/log/postgresql$ free -m
> total used free shared buffers cached
> Mem: 32115 29498 2617 0 110 20128
> -/+ buffers/cache: 9258 22857
> Swap: 0 0 0
>
> ð No swap allowed in this server
>
> Observation:
> ----------------
> The database seems to run well from the 4th dec between 10:20AM to 10:42 AM before falling back suddently to out of memory.
> It switched to recovery mode and restart automatically.
> There were no problem encountered like that before making any change to the checkpoint parameters.
>
> Attached files
> -----------------
> Syslog file and postgresql.log file are attached. I made some clean up in the postgres log file.
>
> Question:
> ------------
> Is it not recommended to change any "checkpoint_xxx" parameters in the postgresql.conf file when the production is running ?
There's no such recommendation. It should be totally fine to change
those parameters. I don't know why you got the OOM error, but I don't
see changing checkpoint settings could cause that.
- Heikki
From | Date | Subject | |
---|---|---|---|
Next Message | Alvaro Herrera | 2014-12-09 14:14:48 | Re: regression, deadlock in high frequency single-row UPDATE |
Previous Message | Alvaro Herrera | 2014-12-09 13:58:41 | Re: BUG #11986: psql uses pager inside Emacs shell buffer (not a terminal) |