I do have statspack report during this situation and what is mentioned
in the metalink note, exactly has been occured to us. Like, enqueue is
one of the top 5 timed events and under the enqueue activity, I see,
TX (transaction Exclusive) and US (Unod Segments) entries.
We have reported this to Oracle Support and they are still revieiwing
the TAR. My question, why oracle doesn't reuse the already online undo
in place of bring more undo segs. online.
Still puzziling.
On 12/29/05, Hemant K Chitale wrote:
>
> Do you have StatsPack snapshots for those times
> [as presented as an example in Note 301432.1] ?
> Or queries on V$SESSION_WAIT ?
>
>
> At 08:41 PM Thursday, The Human Fly wrote:
>
> >Basically, when transaction comes, we have flag called queue for
> >different status and we see that the query is growing heavily for not
> >process status, and when I go and check the alert.log file, exactly at
> >the same time and second, oracle bring undo segs online. We have
> >noticed this situation thrice comparing with undo segs online process
> >and we determined the same.
> >
> >Note:301432.1 Subject: Performance Slowdown During Heavy Undo Segment
> >Onlining
> >
> >We are exactly facing the same problem mentioned in this metalink note
> >and as per Oracle Support advice, we also enable the event 10511, to
> >prvent SMON taking undo segs. offline.
>
>
> Hemant K Chitale
> http://web.singnet.com.sg/~hkchital
>
>
>
--
Best Regards,
Syed Jaffar Hussain
OCP 8i & 9i DBA,
Banque Saudi Fransi,
Saudi Arabia
----------------------------------------------------------------------------------
"Winners don't do different things. They do things differently."
--
http://www.freelists.org/webpage/oracle-l
Received on Thu Dec 29 2005 - 15:03:18 CST
