Message-ID: <4631C825.9000705@Sheeky.Biz>
Date: Fri, 27 Apr 2007 19:23:41 +0930
From: Shane Ambler <pgsql@Sheeky.Biz>
User-Agent: Thunderbird 1.5.0.10 (Macintosh/20070221)
MIME-Version: 1.0
To: Magnus Hagander <magnus@hagander.net>
CC: Andres Retzlaff <andres_retzlaff@hotmail.com>,
	pgsql-performance@postgresql.org
Subject: Re: Usage up to 50% CPU
References: <20070427080653.GB12895@svr2.hagander.net>
	<BAY141-F1845ACAADC609F4311EA888B4F0@phx.gbl>
	<20070427081350.GD12895@svr2.hagander.net>
In-Reply-To: <20070427081350.GD12895@svr2.hagander.net>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Magnus Hagander wrote:
> On Fri, Apr 27, 2007 at 08:10:48AM +0000, Andres Retzlaff wrote:
>> Hi Magnus,
>>
>> in this case each CPU goes up to 50%, giveing me 50% total usage. I was 
>> specting as you say 1 query 100% cpu.
>>
>> Any ideas?
> 
> No. 1 query will only use 100% of *one* CPU, which means 50% total usage.
> You need at least one query per CPU to reach full 100% of the whole system.
> (Actually, you can get slightly above 50% since the query will run on one
> CPU and the OS and autovacuum and bgwriter can run on the other. But it's
> marginally)
> 
> //Magnus

I would think that as you are sitting and watching the cpu usage, your 
query would seem to taking a while to run, leading me to wonder if you 
are getting a full table scan that is causing pg to wait for disk response?

Or are you running a long list of steps that take a while?


-- 

Shane Ambler
pgSQL@Sheeky.Biz

Get Sheeky @ http://Sheeky.Biz