RE: KCL gc element parent latch
Date: Wed, 16 Dec 2009 13:53:14 -0500
Message-ID: <SNT130-ds838E1A47F12F3D9668534A6870_at_phx.gbl>
I've seen these with full table scans going across instances. Tables can be really small but it kills the rac.
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org]
On Behalf Of Rajesh Rao
Sent: Wednesday, December 16, 2009 12:34 PM
To: Tanel Poder
Cc: oracle-l_at_freelists.org
Subject: RE: KCL gc element parent latch
Thank you Tanel.
"so if you have lots of GC ops, these latches will be used. Fix the GC traffic and the latching problem will go away."
So, those waits would be an effect, rather than the cause leading to gc cr request waits. These waits themselves are not significant. We see fast object checkpoints as well as this KCL latch that seems to precede most of the gc cr request waits we see. We know the fix for the gc waits.
Thank you for the script. That goes into the same directory where I have the snapper one stored :-)
Raj
-- http://www.freelists.org/webpage/oracle-lReceived on Wed Dec 16 2009 - 12:53:14 CST