Message-ID: <42EFA55C.3080903@archonet.com>
Date: Tue, 02 Aug 2005 17:54:52 +0100
From: Richard Huxton <dev@archonet.com>
User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317)
MIME-Version: 1.0
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Tobias Brox <tobias@nordicbet.com>,
	pgsql-performance@postgresql.org
Subject: Re: "nice"/low priority Query
References: <20050802160434.GD9278@tobias.lan>
 <11484.1122999570@sss.pgh.pa.us>
In-Reply-To: <11484.1122999570@sss.pgh.pa.us>
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 7bit

Tom Lane wrote:
> Tobias Brox <tobias@nordicbet.com> writes:
> 
>>Is there any ways to give postgresql a hint that a
>>particular SQL call should be run at lower priority?  Since every db
>>connection has a pid, I can manually run "renice" to scheduele it by the OS
>>- but of course I can't do it manually all the time.
> 
> And it won't help you anyway, because renice only affects CPU priority
> not I/O scheduling ... which, by your description, is the real problem.
> 
> I think the only thing that's likely to help much is trying to arrange
> that the "simple" queries only need to touch pages that are already in
> memory.  Some playing around with shared_buffer sizing might help.
> Also, if you're not on PG 8.0.*, an update might help.

Would it be useful to be able to re-use the vacuum_cost_xxx settings in 
8.0 for this sort of thing? I'm thinking a long-running report query 
isn't that different from a vacuum.

--
   Richard Huxton
   Archonet Ltd