Re: CR Block copies, always 6 in 10.2.0.3?

  • From: "K Gopalakrishnan" <kaygopal@xxxxxxxxx>
  • To: ukja.dion@xxxxxxxxx
  • Date: Mon, 24 Mar 2008 21:07:08 -0700

Amonte,

You are right about the BBW timeout of 1sec. But IIRC this was little
different from regular BBW and I will have to go back to my notes to
confirm the 10cs behavior.

Let us not  blame the  RAC or lightwork rule for limiting number  CR
buffers in the buffer cache. In fact there is no hard limit for number
of cr copies in RAC or single instance. This is just a soft limit -
means this limit is enforced  only when there is a memory pressure in
the buffer cache. During that time, upto  6 copies of the CR buffers
are allowed to be part of the hot part  MRU list and the remaining CR
buffers are linked to the Auxiliary list which are ready for expiry.
If your buffer cache is sufficiently large and there is no memory
pressure, we don't need to enforce the number of cr copies limitation.
This is  one of the reasons for not able to simulate this behavior in
the simple lab tests.


> It's a bit vague that i can't 100% understand what the author is trying to
> say.
> Seems that he is referring buffer busy waits wait event but the timeout of
> buffer busy wait is 1s(100cs) not 10cs.
> Asking it to the author seems to be good solution.
>
> Jonathan Lewis has stated a quite probable answers to this situtation.
> My additional guesswork is that Oracle generates more CR blocks than the
> number limited by _db_block_max_cr_dba
> when light work rule is applied on RAC system.
>  But I don't have access to RAC at the moment that can't verify it.
>
>  Dion Cho
>



-- 
Best Regards,
K Gopalakrishnan
Co-Author: Oracle Wait Interface, Oracle Press 2004
http://www.amazon.com/exec/obidos/tg/detail/-/007222729X/

Author: Oracle Database 10g RAC Handbook, Oracle Press 2006
http://www.amazon.com/gp/product/007146509X/
--
//www.freelists.org/webpage/oracle-l

Other related posts:

  1. Home
  2. oracle-l
  3. Archive
  4. 03-2008