[Zope] strange zope problems of the crashing kind

[Leichtman, David J]

Alright Zopers, I'm looking for a hint, a suggestion, just about anything
would help since I can't, for the life of me, figure out what's wrong.

Here's the run-down.
We are running Zope 2.2.0 on Solaris, and we're running it through Apache +
SSL. We're using a PostgreSQL DB with the ZPoPyDA. We are also using
FSSession 0.4.0 for session management.
Previously, we were using Zope 2.1.6, accessing ZServer directly, and using
FSSession 0.3.1.

Here's the problem.
Previously, when we were accessing ZServer directly, Zope would
spontaneously quit. No reason, it would just die. Restarting it would bring
it back up, but minutes, sometimes hours, sometimes days later, it would
spontaneously die again. I asked the list, but no one had had such problems
(except Dieter, but that was an Oracle problem).
So now that we're running it through Apache, we're still, apparently, having
the same problem. What will happen is that every once in a while (more
frequently with more traffic), Zope will hang indefinitely. The only way to
get over it is to restart Zope. So apparently, the ZServer (which I guess
takes requests from the pcgi process), is still going down as before.

The question is, why is this happeneing? Or, more to the point, how can I
figure out why it is happeneing? I looked through the Zope logs and the
Apache logs and found nothing useful. I've been thinking that it might
correspond to a heavy database access. Perhaps there's a problem with the
ZPoPyDA. Or maybe it's FSSession (which, incidently, doesn't seem to be
expiring cookies correctly). Or maybe it's a compatability issue with
Solaris. I really don't kow, and I don't know how to go about finding out.

So any help anybody can give me would be fantastic. If anybody's had any
kind of similar problems, or if anybody has any suggestions, it would be
much appreciated.
I've racked my brains trying to figure it out, and I just can't.

Thanks.

Dave Leichtman
University of Oklahoma