This bug appears to be something deep in the bowels of one of our module's reference counting/memory allocation thingys (or so says gdb). I'm in over my head. Gonna get the big guns in. Might take a little while. In any case, are you both (Sven and Phillip) running Zope 2.3.1? Or is this another version?
i'm using : Zope version: Zope 2.2.2 (source release, python 1.5.2, linux2) Python version: 1.5.2 (#1, Dec 21 2000, 15:29:08) [GCC egcs-2.91.66 19990314/Linux (egcs- System Platform: linux-i386 i'm waiting for the new hardware to be delivered for us to change zope/mysql/linux/hardware/hosting all at the same time ;-)
I have been able to reproduce the bug under 2.3.1 (actually the 2.3 branch of CVS), but I cannot reproduce it under a "trunk" CVS checkout. So it would seem we have a good start on finding out what it is. It would be interesting to find out if you folks are using something other than 2.3.1.
we're using 2.2.2 and i think (but i can't remember) that it did too with the release just before this one (2.2.1 maybe ?) somewhat less often but since this is due to the amount of requests, it has only been visible when traffic rose. it's somewhat unerving since zope dies every 2 hours max. oh, and by the way, sometimes, i have to halt/restart zope to make it work again and sometimes i have to also halt/restart mysql and it works again afterwards ??? weird... $ven