Re: Postgres stats collector showing high disk I/O - Mailing list pgsql-general

From Justin Pasher
Subject Re: Postgres stats collector showing high disk I/O
Date
Msg-id 4BD215F2.307@newmediagateway.com
Whole thread Raw
In response to Re: Postgres stats collector showing high disk I/O  (hubert depesz lubaczewski <depesz@depesz.com>)
Responses Re: Postgres stats collector showing high disk I/O  (Justin Pasher <justinp@newmediagateway.com>)
Re: Postgres stats collector showing high disk I/O  (Alvaro Herrera <alvherre@commandprompt.com>)
Re: Postgres stats collector showing high disk I/O  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
----- Original Message -----
From: hubert depesz lubaczewski <depesz@depesz.com>
Date: Fri, 23 Apr 2010 23:40:35 +0200
Subject: Re: [GENERAL] Postgres stats collector showing high disk I/O
To: Justin Pasher <justinp@newmediagateway.com>
CC: pgsql-general@postgresql.org
> On Fri, Apr 23, 2010 at 03:27:55PM -0500, Justin Pasher wrote:
>
>> haven't tweaked any settings from the defaults. My
>> $PGDATA/global/pgstat.stat file is about 18MB, if that helps. Does
>> it really rewrite this entire file every 500ms? Alvaro suggested
>> resetting the stats, but I'm having trouble figuring out how to do
>> that. Seems like pg_stat_reset() is post- 8.1 ...?
>>
>
> I don't have 8.1 handy, but according to this:
> http://www.postgresql.org/docs/8.1/interactive/monitoring-stats.html
> pg_stat_reset() should be available in 8.1.
>
> Best regards,
>
> depesz
>

Agh... I used pg_stats_reset (with an s) when searching for it. I ran
the function and it returned true, but the stats file only shrunk by
~100k (still over 18MB total). Is there something else I need to do?
Does this mean the file is mostly bloated with bogus data that it can't
"reset"? I'm guessing I should just try to delete the file outright?

--
Justin Pasher

pgsql-general by date:

Previous
From: hubert depesz lubaczewski
Date:
Subject: Re: Postgres stats collector showing high disk I/O
Next
From: Giancarlo Boaron
Date:
Subject: Problem with pg_prepare