Re: latch free - cache buffers chain problem

I'm afraid this may not help.
For indexes, PCTFREE is only of use when creating the index. When a new row arrives, the value determines the leaf block that has to be used - so a block that is in a 'busy area' of the index will eventually get hot irrespective of how empty it starts.

Scott's follow-up is interesting. You've identified the block as being in an index - if you can check that it is the ROOT block, (typically this is the first block after the segment header and any 'freelist group' blocks (if you have multiple freelists groups)) then you may be looking at the same issue.

Of course, as Kyle points out, the issue may be related to the way the block is used - I'm sure I could set up a scenario making use of (or perhaps abusing would be a better word) a 'temporary' table # so that a couple of its index blocks get very hot.

--
Jonathan Lewis
http://www.jlcomp.demon.co.uk

Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html

Author of:
Practical Oracle 8i: Building Efficient Databases

Screen saver or Life saver: http://www.ud.com
Use spare CPU to assist in cancer research.

Gulabchand Jaiswal wrote in message ...


>oraperf_at_yahoo.com (Kyle Hailey) wrote in message

news:<3bd85efa.93575314_at_newsfeeds-goliath.1usenet.net>...


>> But the main cause of waits on the cache buffers chains latch comes

>> from users reading and writing the same block in and inefficient way.

>> If writers don?t commit their changes immediately, then readers who

>..................snip............................

>> Comments appreciated

>>

>> Best Wishes

>> Kyle  Hailey

> oraperf.sourceforge.net

>

>Kyle,

>Thx for an excellent amd insightful response and Thanks to all those who

>responded. We are getting there. The contention is on block in an index !

>We recreated it more pctfree ( thx jonathan) and need to run more tests.

>Will keep you guys posted how results turn out..

>

>Thx again