Re: ORA-00600 [17281] [1001] from JBOSS

I was able to recreate the ORA-00600 on demand using the bad_func example to throw a NO_DATA_FOUND error (but not really throw it), and then wait for JBoss to call the IdleRemover. However the ORA-00600 was only generated when JBoss was using the OJDBC 10.2.0.3 jar. With the 10.2.0.2 jar, no ORA-00600 was generated.

So we'll be reverting to the 10.2.0.2 jars. I can't say if that means the application will stop generating ORA-01403 errors, so right now we're just treating the symptom and not the cause.

I've done a little write-up here:
http://seilerwerks.blogspot.com/2007/04/heart-of-darkness.html

Don.

On 4/10/07, Don Seiler <don_at_seiler.us> wrote:
> Oh I also forgot to mention that every stack trace has an ORA-01403
> "no data found". Coincidentally, this article popped up on my RSS
> reader today:
>
> http://blogs.ittoolbox.com/oracle/guide/archives/minitip-9-no-data-found-bug-or-feature-15602
>
> It talks about how ORA-01403 doesn't get throw when it should be. I'm
> wondering if the applications are causing an 01403, but don't know
> about it, and then perhaps JBOSS is seeing the 00600 when it goes to
> clean up those cursors that had met the 01403 and throws that error.
>
> Of course, that last paragraph is really just loosely connecting
> possibly completely unrelated dots.
>
> Don.
>
> On 4/10/07, Don Seiler <don_at_seiler.us> wrote:
> > I've finally encountered the dreaded ORA-00600. It is popping up from
> > time-to-time since Sunday evening, when we had a new software release
> > (of course). The stack trace always lists JBOSS and our app server as
> > culprits. There are no errors in the JBOSS logs, but one of the
> > developers thinks that JBOSS's IdleRemover calls seem to be in sync
> > with some of the errors, although I don't get an ORA-00600 for every
> > IdleRemover call.
> >
> > Metalink tells me that this particular brand of 00600 happens "while
> > attempting to close all cursors for an instantiation." (which would
> > support the IdleRemover theory) There are some other related bugs as
> > well. However all of them say they have been fixed by 10.2.0.1. My
> > RDBMS is 10.2.0.2 (on 32-bit RHEL 3).
> >
> > One thing that did change on the app server was that we upgraded the
> > ojdbc14.jar from the 10.2.0.2 release to the 10.2.0.3 release. We're
> > looking at reverting to see if that is indeed the culprit, but I'm
> > wondering if anyone has seen any such incompatibilities.

-- 
Don Seiler
http://seilerwerks.blogspot.com
--
http://www.freelists.org/webpage/oracle-l