Re: AIX 4.3.3 / 8.1.7 & Timed Statistics & Tuning Question

Lisa,

   I have had TIMED_STATISTICS set to true on 5 different 8.1.7 databases (4 at 8.1.7.4 (64-bit), one at 8.1.7.0 (32-bit)) on 3 different servers for some time with no perceived problems. These servers are either at AIX maintenance level 9 or 10.

I have struggled with swap space consistently growing until memory is exhausted on one of these servers. It is at AIX 4.3.3 maintenance level 9. There is a memory leak in fclose() at this level that I think is causing the problem. Refer to APAR IY22308.

You don't mention whether the databases are on jfs file systems or not. If so, be aware that the 'fre' column of the 'vmstat' command will typically be small, since some real memory is used as a cache for file system data. (And the virtual memory management page stealer takes a lazy approach, only stealing to keep the free list above the 'minfree' value.)

IBM has a white paper on paging space that may help: http://www-1.ibm.com/servers/aix/products/aixos/whitepapers/pagingspace.pdf

Also, the following commands with give the top 5 users of memory and paging space, respectively:

  svmon -Put 5
  svmon -Pgt 5

Haven't worked with PeopleSoft, so I can't speak to its effect.

Hope this was relevant,
Scott

>>> Lisa.Koivu_at_efairfield.com 2/6/03 9:23:58 AM >>>
Good morning everyone -
Quick poll for those of you on 8.1.7 and AIX 4.3.3: Do you have TIMED_STATISTICS = true? Have you encountered any problems with it? The databases I inherited have this set false all over the place, hence my tuning efforts are really limited. However I don't want to change it without checking around first. And a tuning question:
This environment (peoplesoft) is very very low on memory. When the app servers and databases are up there's less than 50MB of memory free. Adding hardware is not a choice here. The databases have 100MB set for the SGA. It really looks like not much thought went into some of the parm settings. What I've read about tuning says that you must have a goal in mind. Well, afaik nothing is "broken", nothing is suffering - then again, no one really paid much attention to Oracle. It was up, fine, move on. Am I on the wrong path if my goal for tuning is to figure out if I can reduce the size of the SGA and redo logs without adversely affecting performance? Any comments are appreciated. Thanks everyone

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Scott Behrens
  INET: behrens-2_at_medctr.osu.edu

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).