Re: Log file sync and log file parallel write.
Date: Thu, 06 Mar 2008 19:55:31 +0800
Message-Id: <200803061155.m26BtOXk001069@smtp12.singnet.com.sg>
I won't come back with the blind (sometimes too-quick-off-the-mark) question
"why are you rebuilding indexes". There can be valid reasons if you
have experience
to show that a rebuild becomes necessary.
What I want to ask is : Are you rebuilding with NOLOGGING ? Gone are the days when Index Rebuild should cause singificant LGWR activity.
10 index rebuilds if *sequential* should be no different than 1
long running query !
There's only 1 process / 1 thread. Are you using Parallelism
? Maybe the server / IO
cannot handle multiple large reads occurring in parallel ? Would certainly be
a hardware issue if you have had "timeouts" when reading from disk !
Are you using Async I/O and encountering async timeouts ?
When you say "a simple rebuild of 10 indexes", how large are those indexes ? You could guage the total volume of IO being performed.
Last observation : See if you can switch from REBUILD to DROP and CREATE
(or test for a few of these 10 indexes) ? Could be a problem with
the particular
blocks the existing indexes are on ?
Hemant K Chitale
At 03:00 AM Tuesday, Johnson, William L (TEIS) wrote:
>
>A simple exercise of rebuilding about 10 different indexes on-line
>serially takes 45 minutes. s
>Index rebuilds shouldn't be doing this to our system. Our
>production database was crashed yesterday due to other applications
>"timing out" through the cluster while trying to read data while
>this type of activity was occurring.
>
Hemant K Chitale
http://hemantoracledba.blogspot.com
"There is more to life than increasing its speed."
Mohandas Gandhi Quotes
: http://www.brainyquote.com/quotes/authors/m/mohandas_gandhi.html
-- http://www.freelists.org/webpage/oracle-lReceived on Thu Mar 06 2008 - 05:55:31 CST