RE: 9.2 - Parallel Query Not Working

David

            Is this RAC? Can you print exactly values for parallel_instance_group and instance_groups?

From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mark W. Farnham
Sent: Thursday, February 28, 2008 11:17 AM To: david_at_david-aldridge.com; 'Oracle List' Subject: RE: 9.2 - Parallel Query Not Working  

How much of that table is already in the buffer cache when PQ stops being operative?  

mwf  

From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of David Aldridge
Sent: Thursday, February 28, 2008 12:03 PM To: Oracle List
Subject: 9.2 - Parallel Query Not Working  

On a 9.2.0.6 database on Solarix 64, which is created every day from a magical SAN-based mirror process thing (excuse the tech talk), we alter a particular table to a default degree of parallelsim of 12. A process which performs a full scan of that table picks up the degree of 12 and completes in a couple of hours.  

Later in the day we can no longer prompt the instance to give any parallelism.  

I've tried everything obvious ... setting parallel_adaptive_multiuser to false, parallel max servers to 48, used hints, made sure the table's default dop is 12, checked that PROCESSES is reasonably high (500+), that the server is not very busy (less than 10 active sessions), ALTER SESSION FORCE PARALLEL QUERY, but nothing we do will prompt PQ to be used on even the most simple SELECT * FROM ... queries.  

Any thoughts on this are much appreciated.

--
http://www.freelists.org/webpage/oracle-l