Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Rollback waiting on 'cache buffer chains' latch - SOLVED

RE: Rollback waiting on 'cache buffer chains' latch - SOLVED

From: Allen, Brandon <>
Date: Tue, 9 Jan 2007 20:15:06 -0700
Message-ID: <04DDF147ED3A0D42B48A48A18D574C45059E279B@NT15.oneneck.corp>

Thanks everyone for your help. The only solution we could find, along with Oracle Support, was to bounce the instance and luckily it came back up with no problems. We don't have any solid root cause yet, but just a few possibly related factors to keep in mind:

        o We just upgraded to a couple months ago (never had this problem on

        o The same phantom session that was holding these locks also had a deadlock earlier this morning

        o The phantom process was killed earlier with a 'kill -9'

So some combination of these and possibly other factors left us in this undesirable situation. We're done with this for tonight but I'll let you know if we or Oracle Support come up with any other useful information.


-----Original Message-----
From: Andrey Kriushin [] Sent: Tuesday, January 09, 2007 5:57 PM
To: Allen, Brandon
Cc: Bobak, Mark;; oracle-l Subject: RE: Rollback waiting on 'cache buffer chains' latch


I'm afraid there are no good news for you. And I agree with Mark that there is something weird with (latch?) clean up/release. Unfortunately, it's not that easy to catch. And usually the only working drug is to bounce the instance.

And again I'm late - sorry, its about 4 am here is Moscow. Hopefully those guys from OSS asked you to take a couple of systemstate dumps or even "latches" dumps before you managed to bounce the instance. Sorry, the idea of LATCHES dumps came to me too late. It was a chance to see the holder, if the latch was kept permanently (due to failed cleanup or other bug of that sort in another session).

Definitely all that looks like a bug. So be prepared and arm yourself with a list of dumps to take on next occurence. Hopefully, it would not be required then.

Also it is a good idea to ask Tanel or other fellow gurus on the list for a piece of code to monitor the particular latch structure directly in SGA.

PS. Could you, please, share the results of the investigations with the list when it will be resolved? We have a bunch of the customers with critical 24x7 databases which are already on or have plans to patchset-up (is it correct wording?) their production databases to this release.

On Wed, January 10, 2007 02:47, Allen, Brandon wrote:
> Any ideas on how to clear it other than bouncing the instance?

Privileged/Confidential Information may be contained in this message or attachments hereto. Please advise immediately if you or your employer do not consent to Internet email for messages of this kind. Opinions, conclusions and other information in this message that do not relate to the official business of this company shall be understood as neither given nor endorsed by it.

Received on Tue Jan 09 2007 - 21:15:06 CST

Original text of this message