Oracle 11.2.0.2
Solaris 10 64-bit on Oracle T5
EMC storage (fast cache enabled)
Last Wednesday night we moved a production database to new hardware. We do
not have the option to move it back. On Friday we started experiencing
severe performance issues. ASH reports show the "SELECT...FOR UPDATE
WAIT..." running as the top SQL for events and row sources. This is NOT
new code. What can I look at, is there anything I can do, short of
killilng sessions, to get this under control? I'm opening a ticket with
Oracle, but this forum generally responds faster.
I'm not really familiar with how the SELECT FOR UPDATE WAIT works and
haven't found a good explanation yet. The select is on a partitioned
table. Does it lock only a row or does it lock the partition? At times I
can see a dozen or more of these statements running. Once they complete, I
then see a lot of waits on log file sync.
If anyone knows, how does having fast-cache enabled affect this?
Thanks for any suggestions/guidance.
--
Sandy
GHX
