stats collector process: WRDSK = WRDSK_CANCEL?

Lists: pgsql-general
From: Лев Ласкин <laskin82(at)gmail(dot)com>
To: pgsql-general(at)postgresql(dot)org
Subject: stats collector process: WRDSK = WRDSK_CANCEL?
Date: 2009-12-22 09:30:05
Message-ID: 4B30919D.9030903@gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-general

Hi.
We have installed Postgresql 8.3.3. When we see the output of atop-d, we
see just such a string

db0: ~ # atop 15
ATOP - db0 2009/12/22 12:14:29 15 seconds elapsed
DSK | sda | busy 50% | read 72 | write 207 | avio 27 ms |

PID RDDSK WRDSK WRDSK_CANCEL DSK CMD 1 / 2
11296 0K 185.0M 185.0M 97% postgres

db0: ~ # ps auxww | grep postgre
postgres 11296 4.3 0.1 24244 13028? Ss Nov24 1734:25 postgres: stats
collector process

Why WRDSK = WRDSK_CANCEL? So it should be? Perhaps we were wrong in the
settings.

Who can help anything in this matter?


From: Magnus Hagander <magnus(at)hagander(dot)net>
To: Лев Ласкин <laskin82(at)gmail(dot)com>
Cc: pgsql-general <pgsql-general(at)postgresql(dot)org>
Subject: Re: stats collector process: WRDSK = WRDSK_CANCEL?
Date: 2009-12-22 18:40:56
Message-ID: 9837222c0912221040m707c2525pd6015b56c95a87a2@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-general

On Tue, Dec 22, 2009 at 10:30, Лев Ласкин <laskin82(at)gmail(dot)com> wrote:
> Hi.
> We have installed Postgresql 8.3.3. When we see the output of atop-d, we see
> just such a string
>
> db0: ~ # atop 15
> ATOP - db0 2009/12/22 12:14:29 15 seconds elapsed
> DSK | sda | busy 50% | read 72 | write 207 | avio 27 ms |
>
>   PID RDDSK WRDSK WRDSK_CANCEL DSK CMD 1 / 2
> 11296 0K 185.0M 185.0M 97% postgres
>
> db0: ~ # ps auxww | grep postgre
> postgres 11296 4.3 0.1 24244 13028? Ss Nov24 1734:25 postgres: stats
> collector process
>
> Why WRDSK = WRDSK_CANCEL? So it should be? Perhaps we were wrong in the
> settings.
>
> Who can help anything in this matter?

WRDSK_CANCEL is data that was written, but deleted before it was
flushed to disk. That is normal behaviour for the stats temp file,
which is written and deleted within half a second.

8.4 will do a lot less writes here, but on 8.4 it's perfectly normal.

--
Magnus Hagander
Me: http://www.hagander.net/
Work: http://www.redpill-linpro.com/


From: Magnus Hagander <magnus(at)hagander(dot)net>
To: lls <laskin82(at)gmail(dot)com>
Cc: pgsql-general <pgsql-general(at)postgresql(dot)org>
Subject: Re: stats collector process: WRDSK = WRDSK_CANCEL?
Date: 2010-01-04 08:29:58
Message-ID: 9837222c1001040029x121d4b38sa5938846a4d12860@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-general

On Mon, Jan 4, 2010 at 06:28, lls <laskin82(at)gmail(dot)com> wrote:
> 22.12.2009 21:40, Magnus Hagander пишет:
>
> On Tue, Dec 22, 2009 at 10:30, Лев Ласкин <laskin82(at)gmail(dot)com> wrote:
>
>
> Hi.
> We have installed Postgresql 8.3.3. When we see the output of atop-d, we see
> just such a string
>
> db0: ~ # atop 15
> ATOP - db0 2009/12/22 12:14:29 15 seconds elapsed
> DSK | sda | busy 50% | read 72 | write 207 | avio 27 ms |
>
>   PID RDDSK WRDSK WRDSK_CANCEL DSK CMD 1 / 2
> 11296 0K 185.0M 185.0M 97% postgres
>
> db0: ~ # ps auxww | grep postgre
> postgres 11296 4.3 0.1 24244 13028? Ss Nov24 1734:25 postgres: stats
> collector process
>
> Why WRDSK = WRDSK_CANCEL? So it should be? Perhaps we were wrong in the
> settings.
>
> Who can help anything in this matter?
>
>
> WRDSK_CANCEL is data that was written, but deleted before it was
> flushed to disk. That is normal behaviour for the stats temp file,
> which is written and deleted within half a second.
>
> 8.4 will do a lot less writes here, but on 8.4 it's perfectly normal.
>
>
>
>
> Thank you, Magnus. How about 8.3.9? Should I switch to this version to
> reduce writes?

You should switch to this version, yes, because it contains important
bugfixes. But it will not change the write behaviour.

--
Magnus Hagander
Me: http://www.hagander.net/
Work: http://www.redpill-linpro.com/