Re: Using RDBMS 10.2.0.3: Have You Experienced Enq: Hwm - Contention?
From: <Laimutis.Nedzinskas_at_seb.lt>
Date: Wed, 6 Apr 2011 09:10:29 +0300
Message-ID: <OF000155E5.6BCB9B33-ONC225786A.0021685D-C225786A.0021EB7E_at_seb.lt>
our solution was a patch for 6376915 (for oracle version 10204) It helped.
Date: Wed, 6 Apr 2011 09:10:29 +0300
Message-ID: <OF000155E5.6BCB9B33-ONC225786A.0021685D-C225786A.0021EB7E_at_seb.lt>
our solution was a patch for 6376915 (for oracle version 10204) It helped.
After that another LOB event (sorry, I've forgot the details) poped-out but for that we knew what to do - LOB cache helped (thanks to small LOB size.)
Brgds, Laimis N
Please consider the environment before printing this e-mail
From: fmhabash <fmhabash_at_gmail.com> To: Oracle L <oracle-l_at_freelists.org> Date: 2011.04.05 22:52 Subject: Using RDBMS 10.2.0.3: Have You Experienced Enq: Hwm - Contention?
We go hit by this event recently and would like validate the scenario
and the proposed solutions against your experience if you encountered
it. This event affected a LOB segment in an ASSM tablespaces in a RAC
database.
This is what I have found so far ...
1) As temporary workaround, manually add extra space to the LOB segment.
2) Bug 6376915 (affecting 10.2.0.3)
- >= 10.2.0.4: EVENT="44951 TRACE NAME CONTEXT FOREVER, LEVEL < 1 - 1024 >"
- set PCTVERSION to 100: but never able to use blocks freed by updated blobs 3) Bug 6413373 (LOB in RAC and ASSM) - Lob HW enqueue contention: LOBs in ASSM tablespaces
- Set FREEPOOLS is set to a value >= the number of instances
- Partition the table into a number of partitions sufficient 4) Try alter system set"_insert_enable_hwm_brokered"=false
If you did encounter this issue, what was you rca and final resolution?
Thank you.
-- http://www.freelists.org/webpage/oracle-l -- http://www.freelists.org/webpage/oracle-lReceived on Wed Apr 06 2011 - 01:10:29 CDT