RE: Checkpoint locks
Date: Tue, 17 Jun 2008 21:04:08 +0800
Just one hypothesis:
As you didn't have TEMP tablespace, the SQL workareas for some user sessions were allocated as temporary segments in SYSTEM tablespace.
In this case whenever a workarea was released, its temporary segment was dropped - which in turn caused an object checkpoint on the segment, which in turn caused CKPT process to take the KO enqueue on that segment.
On the other hand you may have just hit some bug or issue with (parallel) rollback of a transaction...
At this point I realized that I had forgotten to create tempfiles and did
it. Once temp files were created
the app servers started and we were able to connect. This happened just after we cloned our prod to test and when trying to app servers. Oracle 10.2.
Would checkpoint process need any space for sorting? I thought its job was to flush the buffers to the data files.
http://www.freelists.org/webpage/oracle-l Received on Tue Jun 17 2008 - 08:04:08 CDT