I am having serious stability issues with Zope. I brought this up about a month ago and, based on very helpful advice, succeeded at setting up Apache in front of Zope. I was assured this would greatly increase stability by filtering out script-kiddie attacks. Alas, it hasn't. Zope still craps out at random, once a day or so. Might the malformed requests be coming in through port 8080 directly, crashing Zope that way? I can't shut off the port in Zope or Apache won't be able to communicate. However, I might be able to set up a packet filter that will stop traffic on port 8080 on the external NIC... for example. But I don't want to plow my way through a kludgey solution that does not address the actual source of the failures. In short, I don't know why I can't trust Zope. This is Bad News because I am trying to finish my first paid Zope contract and I can't deliver an unreliable product. Since Zope is extremely dependable to most people, I am apparently doing Something Wrong. Can anybody tell me if there is any particular problem with my setup? I'm running OpenBSD 3.4 on fairly basic machines: one an HP Pentium III, the other a no-name Athlon box. The operating system seems to have no problem with the hardware. On these I have installed Python 2.1.3 and Zope 2.6.1 and MySQL 3.23 and so on, all from OpenBSD packages. I compiled Apache 2 from source with mod_proxy, mod_http_proxy and mod_rewrite and it successfully routes content from Zope, but was returning Bad Gateway messages until I set it to ignore errors (is this normal? nobody mentioned it). My strategies from here: (1) Try another OS. I put FreeBSD on one machine, and although I did not get the whole thing running it appears that the software in the ports collection is the same as OpenBSD so I don't anticipate any benefit. I would like to hear OS suggestions... except those made by Microsoft. Would some flavour of Linux make my life any easier? (2) Upgrade the software. This will mean building from source and facing, if everything goes perfectly smoothly, hours and hours of compiling time on a P III. Again, this has no promise of helping and could turn into a big waste of time if they don't compile happily. Any help is appreciated! TIA, Murray