Re: help tuning queries on large database

From: Ron
Subject: Re: help tuning queries on large database
Date: ,
Msg-id: 7.0.1.0.2.20060111005534.0378a630@earthlink.net
(view: Whole thread, Raw)
In response to: Re: help tuning queries on large database  (Mark Lewis)
List: pgsql-performance

Tree view

help tuning queries on large database  (peter royal, )
 Re: help tuning queries on large database  (Tom Lane, )
  Re: help tuning queries on large database  (David Lang, )
 Re: help tuning queries on large database  (Harry Jackson, )
 Re: help tuning queries on large database  ("Luke Lonergan", )
  Re: help tuning queries on large database  (peter royal, )
   Re: help tuning queries on large database  ("Luke Lonergan", )
    Re: help tuning queries on large database  (peter royal, )
     Re: help tuning queries on large database  ("Luke Lonergan", )
   Re: help tuning queries on large database  (Ron, )
 Re: help tuning queries on large database  (Ron, )
  Re: help tuning queries on large database  (Kelly Burkhart, )
   Re: help tuning queries on large database  (Harry Jackson, )
  Re: help tuning queries on large database  (Mark Lewis, )
   Re: help tuning queries on large database  (Ron, )

At 07:28 PM 1/10/2006, Mark Lewis wrote:
>Ron,
>
>A few days back you mentioned:
>
> > Upgrade your kernel to at least  2.6.12
> > There's a known issue with earlier versions of the 2.6.x kernel and
> > 64b CPUs like the Opteron.  See kernel.org for details.
> >
>
>I did some searching and couldn't find any obvious mention of this issue
>(I gave up after searching through the first few hundred instances of
>"64" in the 2.6.12 changelog).
>
>Would you mind being a little more specific about which issue you're
>talking about?  We're about to deploy some new 16GB RAM Opteron DB
>servers and I'd like to check and make sure RH backported whatever the
>fix was to their current RHEL4 kernel.
There are 3 issues I know about in general:
1= As Peter Royal noted on this list, pre 12 versions of 2.6.x have
problems with RAM of >= 4GB.

2= Pre 12 versions on 2.6.x when running A64 or Xeon 64b SMP seem to
be susceptible to "context switch storms".

3= Physical and memory IO is considerably improved in the later
versions of 2.6.x compared to 2.6.11 or earlier.

Talk to a real Linux kernel guru (I am not) for details and specifics.
Ron




pgsql-performance by date:

From: Michael Fuhr
Date:
Subject: Re: Index isn't used during a join.
From: Simon Riggs
Date:
Subject: Re: NOT LIKE much faster than LIKE?