Re: Shared Pool bug in 7.2.2?

Portland State has also had the same problem, we are on a Sequent UNIX PTX 2.1, running 7.2.2.3 and had the same problems when a about 160 users started to bang away,. We also upped our SP from 9M to 33M, helped but still had same error. We have not received the 7.2.3 fix yet but were given the actual ora library patches and had to recompile/link oracle. Currently the problem seems fixed, we have heavy loads today (240+ users) all banging away and so far the problem "seems" fixed. When I had b/estat running during the problem my Latch - Shared pool hit_ratio was in the dumpster (below .4) and now I am back into the .93+ area.
>From what I was able to get from oracle (don't quote me :-) ) there was
a problem with the LRU algorithum on the Shared Pool not releasing properly and causing fragmentation, flushing the shared pool was the only solution until this patch.
I sure hope that 7.2.3 works - if not then I'm not even going to waste my time. Hope this helps.
Dave

Dave DeVore           Project Manager / DBA               devored_at_pdx.edu
Portland State University                                    Portland, OR
Office of Information Technologies                           503-725-3112

On Mon, 8 Jan 1996, Sam Siciliano wrote:

> Date: Mon, 8 Jan 1996 13:21:21 -0800
> From: Sam Siciliano <Sam_Siciliano_at_PGN.COM>
> To: Multiple recipients of list ORACLE-L <ORACLE-L_at_SBCCVM.BITNET>
> Subject: Shared Pool bug in 7.2.2?
>
> We have a new application, a package on AIX 3.2.5
> and Oracle 7.2.3, coming up, and with only one or two
> people on banging in some table entries,
> administration, etc, we are getting the ora-04031
> message, out of shared memory. The canned answer
> for this problem number is to increase the size of the
> shared_pool. I have gone from 3,000,000 to 9,000,000,
> and they still have the problem. When we bounce
> Oracle, then it works fine.
>
> I called Oracle support, and they said they didn't know
> of problems in 7.2.2, but then the fax they sent me
> warned of a bug 318582 which is supposed to be fixed
> in 7.2.3. Has anyone run across this bug or a similar
> problem as mine?
>
> Again, since this is a new application, I don't know if
> the problem is unique to 7.2.2.3 or even if the
> application has bad code. However, this is a package
> running out on a lot of sites, so I am suspecting Oracle
> more than the application vendor. It appears that
> memory is not getting freed up or is getting
> fragmented, so that eventually we get the out of
> memory message. I tried issue the alter system
> command to flush the shared pool, and that gets us
> going again. Any body seen anyting like this in 7.2.2.3
> and know of any fix?
>
Received on Mon Jan 08 1996 - 17:12:09 CST