Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle8i utlbstat/utlestat Solaris 2.6

Re: Oracle8i utlbstat/utlestat Solaris 2.6

From: Martin Haltmayer <Martin.Haltmayer_at_0800-einwahl.de>
Date: Fri, 24 Dec 1999 18:01:05 +0200
Message-ID: <386398C1.4D9E6EFC@0800-einwahl.de>


Hello Steve,

many thanks for your enlightenment. Can you please tell me what are the thresholds/ratios/values of 'free buffer waits' and 'write complete waits' to look for? I would have expected more of a hint in the Oracle supplied scripts themselves.

Martin

Steve Adams wrote:
>
> Hello Martin,
>
> The DBWn algorithm has been completely rewritten for 8i. Instead of
> just a write queue and a ping queue there are now lots of queues for
> different types of writes and different priorities and quotas for
> each. Also, instead of DBWn responding to a variety of different types
> of make free request and cross instance calls, there is now a unified
> entry point to DBWn services. So the old 'write requests' statistic is
> no longer meaningful.
>
> Maybe you would do better looking at 'free buffer waits' and 'write
> complete waits' if any, and basing your tuning on those.
>
> Regards,
> Steve Adams
> http://www.ixora.com.au/
> http://www.oreilly.com/catalog/orinternals/
> http://www.christianity.com.au/
>
> On Thu, 23 Dec 1999 23:47:21 +0200, Martin Haltmayer
> <Martin.Haltmayer_at_0800-einwahl.de> wrote:
>
> >Hi all,
> >
> >I tried the utlbstat/utlestat scripts which I knew from Oracle 7.3.4
> >(where they worked fine). But in Oracle 8.1.5 on Sun Solaris 2.6 the
> >"average write queue length" does not give any value (after the query in
> >utlestat it says "0 rows selected"). I looked into the query and it
> >looks for two entries in V$SYSSTAT, the ones with the name = 'summed
> >dirty queue length' and 'write requests'.
> >
> >'write requests' ceased to exist obviously. Also the docs (on NT 4.0) in
> >"Oracle 8i Reference", table 3-13 do not mention it any more.
> >
> >Which is the appropriate statistic name to use?
> >
> >Any help appreciated.
> >
> >Martin
Received on Fri Dec 24 1999 - 10:01:05 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US