Path: news.easynews.com!newsfeed1.easynews.com!easynews.com!easynews!newsfeed.news2me.com!canoe.uoregon.edu!logbridge.uoregon.edu!newsfeed.stanford.edu!postnews1.google.com!not-for-mail
From: raghav_vinjamuri@hotmail.com (Raghav Vinjamuri)
Newsgroups: comp.databases.oracle.server
Subject: Re: PL/SQL:   -- pkg runtime Size Limitations?
Date: 10 Oct 2002 16:54:54 -0700
Organization: http://groups.google.com/
Lines: 58
Message-ID: <2e2fc08e.0210101554.107ed9e@posting.google.com>
References: <2e2fc08e.0210101032.75d082cf@posting.google.com> <3da5cd6b.4654532@news.mobilixnet.dk>
NNTP-Posting-Host: 128.107.253.37
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
X-Trace: posting.google.com 1034294094 3985 127.0.0.1 (10 Oct 2002 23:54:54 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: 10 Oct 2002 23:54:54 GMT
Xref: newsfeed1.easynews.com comp.databases.oracle.server:163804
X-Received-Date: Thu, 10 Oct 2002 16:54:50 MST (news.easynews.com)

kennethkoenraadt@no-spam.hotmail.com wrote in message news:<3da5cd6b.4654532@news.mobilixnet.dk>...
> Hi Raghav,
> 
> I don't see anything suggesting that your package is causing the
> problems, especially since the db crashes. What makes you so sure that
> the pacage is the sinner?
> 
> What does the alert log say when you get ORA-3113 ? And when the db
> crashes ?
> 
> Platform, version ?
> 
> 
> - Kenneth Koenraadt
> 

Hi Kenneth,
   Thanks for your reply.

   I'd want to agree that the 3000 line package
   is fine.

   These are our observations:
   1) 299,999 cases of 300,000 cases works correctly.
   2) Only 1 case (that too when run standalone) causes
      ORA-3113 exceptions.
      We get ORA-7445 errors with a SEGV error.
      e.g.
      ORA-07445: exception encountered: core dump [kocgpn2()+2660]
[SIGSEGV] [Address not mapped to object] [0xFFFFFFF0] [] []
   3) When this 1 case is processed amongst a batch of
      cases (-- not in the first place of the batch),
      then that goes thru fine.   

   Based on the above observations:

   a) Obviously, the data is fine - as the processing
      could not have gone thru - when this case is 
      included in a batch of cases.
   b) The package is also fine - as everything works
      when in the batch mode.

   And, considering that the problem is being
   seen - on account of the "application", there is
   something about the data - that is causing the 
   schizophrenic tendencies to surface :)
   And, we've been thru the data repeatedly, but .. :)
   
   Therefore, the thinking is that there is some 
   memory limit being hit - something to do with
   memory init/utilization in PLSQL pkg variables/collections.

   On a related note, is there any way to monitor
   runtime PLSQL memory consumption for pkg variables
   and collections?   

Thanks again,
-Raghav