Ant: Re: Cancel of a long-running, latch-holding query takes long (8i) - why?
- From: Peter Alteheld <palteheld@xxxxxxxx>
- To: Jared Still <jkstill@xxxxxxxxx>
- Date: Thu, 30 Jun 2005 09:34:00 +0200 (CEST)
Thank you, Jared. I haven't thought of that view! But unfortunately it shows
only the last db_file_scattered_read event before the cancel operation with
rising 'seconds_in_wait'-column value.
In the meanwhile our tar was pointed to bugs about spinning processes when
releasing child chain buffer cache latches. But in those cases the only
workaround was a db-restart while in our case the cancel operation process just
runs long.
Maybe the cancel operation of such a killer query isn't solved too well in 8i.
Peter
Jared Still <jkstill@xxxxxxxxx> wrote:
querying v$session_wait while waiting might offer up some clues.
Jared
On 6/29/05, Peter Alteheld <palteheld@xxxxxxxx> wrote:Hi List,
we had a cancelled query that held 'cache buffer chain' latches on some objects
and took a lot of cpu consumption. The time for the cancel operation takes in
our production db (8.1.7) about an hour, on a dev db (8.1.7) 30 min and on an
upgraded db (10.1.0.2) only some seconds. While we are not able to upgrade
production very soon, I am trying to figure out a solution for 8i.
---------------------------------
Gesendet von Yahoo! Mail - Jetzt mit 1GB kostenlosem Speicher
---------------------------------
Gesendet von Yahoo! Mail - Jetzt mit 1GB kostenlosem Speicher
Other related posts:
- » Ant: Re: Cancel of a long-running, latch-holding query takes long (8i) - why?