Hallelujah! I no longer feel alone in the world. Here is the desperate cry for help which I posted to the world several weeks ago. The thread contains two further messages, both from myself.
http://lists.zope.org/pipermail/zope/1999-December/015473.html
However, no solution was ever found, and I never discovered anyone else who could even empathize with my plight. Until now. :-)
Since posting my original messages, I modified our z2.py file to constrain our Zope installation to a single thread. Since then the problem has not occurred, but I don't consider that to be conclusive, nor do I consider my workaround to be satisfactory.
Please bear with me.I may be completely off track, but after running Zope in several modalities without problems for months on an end (NT4 several versions), we ran into the same problems last week: a request going out to the server, but the server hanged. A Zope restart showed the same after a few minutes. A system restart also. I have never had this problem before, except for once. At that time, as well as at this time, the only thing that helped was (don't laugh) renaming the log file (type: stupid file logger) which had become _very_ large. Now, my http server and thread interaction knowledge is too minimal to be able to explain this, but could a deadlock on the log file lead to log file corruption, and perhaps even in a server deadlock (because the corrupted log file can't be opened). Or is the corruption just a result from the hanging server? Or could it be two threads trying to open/write to the same file (logfile in this case) at once (and the sheer size of the file stand in the way of that leading to an action none of the threads could conclude)??? Has anyone seen this before? This may be complete nonsense, but whatever it may be, renaming/removing the logfile solved the problem for me twice. Oh well, just guessing at the end of a millennium ... Happy new year Rik