Re: Multi-Threaded Server

From: Richard Cronau <saeinc_at_haven.ios.com>
Date: 1996/07/14
Message-ID: <4saum9$mh4_at_news2.ios.com>#1/1


In article <4s43dg$2ah6_at_elmo.cadvision.com>, beatonb_at_cadvision.com says...
>
>In article <4rui7j$1bv_at_news.inforamp.net>,
> rob.bezaire_at_canrem.com (Rob Bezaire) writes:
>> We are running an Oracle 7.2 server on a DEC ALPHA 2100 in single
>> ...
>> We are experiencing frequent ORA-04031 errors which indicate that SGA
>>private memory has been exhausted. There are only a handful of users on the
>>system and the SGA size is set to 200M. Someone suggested using
 mutithreaded
>>server configuration to solve this problem but I haven't read any
>>documentation that confirms this. Does anyone have some advice?
>
>When we had these problems last December & January with 7.2.2.2 thru 7.2.2.4,
>ORACLE had us do the following:
>
> 1. Periodically (frequency depends on your shop's job mix) issue
> 'alter system flush shared pool;'
> We find that on our heavy use instance, we have to issue this
 entry
> every 8 minutes, to ensure that we have less than 1 4031 error
> every 45 days. We are a 7x24 shop.
> 2. Pin ALL ORACLE procedures (like DIANA, etc) into the shared pool
> at instance startup.
> 3. Eliminate ANY and ALL use of MTS.
>
>These approaches work for us, but may not be appropriate for shops with a
>drastically different job mix &/or user base.
>
>We are looking at the feasibility of upgrading to 7.3.2, because it appears
>that the 2 problems (memory leak in SGA, and MTS lockups) have been
>corrected.
>
>Good luck,
>Bill

Do you know if 7.2.3 solves these problems? I know of a project that may suffer as a result of these problems and want to forewarn them prior to these problems occuring.

TIA,

-- 
==============================================================================
=====
Richard J Cronau
Principle
System Architects & Engineers, Inc.
saeinc_at_haven.ios.com
"Helping make client/server work for you"
==============================================================================
=====
Received on Sun Jul 14 1996 - 00:00:00 CEST

Original text of this message