RE: RMAN hangs after 10.2 upgrade

Jared,  

You're right... would have been better if we had done a 10046 trace.  

The control files are ASM on a SAN. We saw the waits on all 6 databases that were upgraded to 10.2.0.2 at the same time. Yes, they're all on the same SAN. There were no database files moved nor were there any SAN changes to my knowledge.  

While we can't definitively identify the 10.2.0.2 as the cause... we still suspect that it is. However, as you say we can't rule out other possibilities.  

Thanks for your help and advice!    

Orysia      

Orysia Husak

Sr. Oracle DBA - Classroom Applications Hosting

University of Phoenix/ Apollo Group, Inc.

Office: 602-546-6934

Mobile: 602-377-8586

orysia.husak_at_apollogrp.edu

From: Jared Still [mailto:jkstill_at_gmail.com] Sent: Friday, March 02, 2007 2:23 PM
To: Orysia Husak
Cc: oramail_at_guidancetech.com; oracle-l_at_freelists.org Subject: Re: RMAN hangs after 10.2 upgrade  

On 3/2/07, Orysia Husak <Orysia.Husak_at_apollogrp.edu> wrote:

        Yes, the control files were recreated in the same locations.          

What comes to mind is while the control files may have been recreated in the same logical location, it may not have been the same physical location as previous.

Were any database files moved?

Is the database on a SAN? Any changes to that at the same time?

Something was causing control file sequential read waits, and then the control files are recreated and the waits disappear.

It is impossible to say after the fact if this was directly attributable

to the migration to 10.2.0.2.

For instance, the problem may have been mitigated by simply copying the old control files to new copies, deleting the originals and renaming the new ones.

Chalk it up to experience.
Next time you encounter something like that (and don't have a 5 minute deadline ) try experimenting a bit and you may come up with a reasonable cause.

Oracle isn't always to blame.

I realize you are the OP on this thread, but the situations are similar.

Now if there were a 10046 trace, that might go a long way towards pinpointing the problem.

--

Jared Still
Certifiable Oracle DBA and Part Time Perl Evangelist

--

http://www.freelists.org/webpage/oracle-l Received on Fri Mar 02 2007 - 15:39:46 CST