Re: Background Writer and performances

From: DANTE Alexandra <Alexandra(dot)Dante(at)bull(dot)net>
To: Martijn van Oosterhout <kleptog(at)svana(dot)org>
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: Background Writer and performances
Date: 2006-07-10 15:46:51
Message-ID: 44B2766B.1080505@bull.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

These graphs represent traffic to the disks, and have been generated
from a "home-made" tool based on "top", "vmstat" and "iostat".
Only PostgreSQL accesses to them, a JVM is launched via BenchmarkSQL but
does not access to the disks on which are stored the data.
BenchmarkSQL stores its reports on the "sdr" disk.

Concerning the logs of the postmaster, I let the defaults values, so I
do not have the queries, timing, statements, ...

The fsync is activated and the "wal_method_fsync" is "fsync" (by default).
Is there a way to be more aggressive with pushing pages out to disk via
PostgreSQL ?

Regards,
Alexandra

Martijn van Oosterhout wrote:

>On Mon, Jul 10, 2006 at 05:06:56PM +0200, DANTE Alexandra wrote:
>
>
>>Hello List,
>>
>>I have uploaded charts on a ftp server.
>>You can access to these 6 graphs by doing
>>ftp visibull.frec.bull.fr
>>
>>
>
>Or more easily, by putting this in your web-browser:
>
>ftp://visibull.frec.bull.fr/PGS_bgwriter/
>
>I'm presuming these graphs are traffic to the disks, right? Do you have
>a measurement of the requests from postgres? I'm not sure how you'd get
>that but it'd probably help with understanding the graphs.
>
>Maybe there's a way to get the kernel to be more aggressive with
>pushing pages out to disk? It has a bgwriter too...
>
>Have a nice day,
>
>

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Scott Marlowe 2006-07-10 15:50:26 Re: VACUUM FULL versus CLUSTER ON
Previous Message Martijn van Oosterhout 2006-07-10 15:29:17 Re: Background Writer and performances