A very LARGE large pool

From: jason arneil <jason.arneil_at_gmail.com>
Date: Thu, 18 Mar 2010 21:00:59 +0000
Message-ID: <fda4898f1003181400s50b1e38chf68e96aa6693ad8_at_mail.gmail.com>


I'm seeing on a RAC instance (linux x86-64) an absolutely enormous large_pool. It's around 2.3GB.

I'm finding it very hard to actually ascertain what sessions are actually holding large quantities of this memory. Is anyone aware of a technique for mapping large_pool allocations to the actual sessions that are grabbing them?

From circumstantial evidence, we think the memory is allocated to sessions from a PRO C application. From my understanding, there are very limited reasons for memory to be allocated from the large_pool, and I don't understand why this application should be grabbing its memory from the large_pool.

Anyone else seen PRO C apps grab large chunks of large_pool?

select pool, name, bytes/(1024*1024) from V$sgastat where pool='large pool';

POOL NAME BYTES/(1024*1024)
------------ -------------------------- -----------------

large pool   session heap		       2040.32632
large pool   ASM map operations hashta		     .375
large pool   free memory			214.27253
large pool   PX msg pool		       1.02615356

on another node of the RAC cluster, (where this particular app is NOT connected) there is hardly any large_pool memory allocation.

Thanks for any suggestions.




http://www.freelists.org/webpage/oracle-l Received on Thu Mar 18 2010 - 16:00:59 CDT

Original text of this message