Re: intermittent very high waits in LGWR on Linux?

From: DA Morgan <damorgan_at_psoug.org>
Date: Fri, 01 Jul 2005 17:22:35 -0700
Message-ID: <1120263761.749846@yasure>

Joel Garry wrote:

>>It could also well be in archivelog mode,

>
> which will slow you down periodically on a single >disk system.
>
> So compare the archiving log times in the alert log with the slowdown
> wallclock times...
>
> I've also seen where cpu's get spun by other things (classically by
> SMON or PMON) while doing various sorts of maintenance, noticeably
> slowing things down for everyone.
>
> jg
> --
> @home.com is bogus.
> Throw out the internet and start all over again.
> http://www.wired.com/news/infostructure/0,1377,68004,00.html?tw=wn_6techhead

Follow-up:

Here's more information on what I am seeing?

The ASM is showing 100% CPU on asm_dbw0_+ASM. No disk activity, it appears that ASM is “doing something”, Looking into trace and alert logs for +ASM show nothing.

In the meantime, that database using this ASM instance is “on hold” with “free buffer waits” holding up all sessions and SQL. The duration that ASM “is lost” is always the same (8 minutes approx.) and the happens cyclically every 8 minutes (approx.).

ASM seems to “wake up” or finish whatever had the CPU pegged, and the waits are gone and database springs back to life.

It appears that ASM takes a nap and the database sees this as a “free buffer wait” situation.

What is going on with ASM? Does the SGA for ASM need to be larger? How can on diagnose the ASM problem? This is being investigated. But any feedback from others will be greatly appreciated.

-- 
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)

Received on Fri Jul 01 2005 - 19:22:35 CDT