[Zope] zope crashing unexplicably, also not releasing ports

David McCuskey dmccuskey@bigfoot.com
Tue, 19 Sep 2000 14:02:08 -0700


i've been using zope for about 6 months now and have had this re-
occurring problem with it randomly crashing.  *worse part about it is
that the port isn't released when it crashes*. this means we can't use
the same startup script to get zope back up, can't make any cron jobs to
check for availability , and must do it manually.

it crashes about every 1-10 days and needs to be restarted on a different
port. for example, just yesterday the machines were rebooted, and today (
after approx 11,000 hits ) it crashed. i checked the logfile, but nothing
out of the ordinary was in there. i even put in the last couple of
entries to see if there was a problem.

i thought it started happening when i moved the data from MySQL to
Oracle, but it could have been that it was only being used by me during
the development. so take this with a grain of salt.

my dtml code is very plain - whatever someone learns in a week of pouring
over the docs that are available. nothing fancy at all.

== Question
has anyone had experience with this problem?

at some point i'd like to get apache running on the servers for some
other cgi stuff. should i just start getting apache configured for zope
and forget the zserver? would this be a more stable environment?


ps, system and traceback included below

== System
Sun Enterprise 450
multiple processors
SunOS 5.7
lots of RAM

Zope 2.1.6
ZOracleDA 2.1.0
Oracle 8i
using 8i setup file

== Traceback

mclaren 11 % ./start -P 9500 &
[1] 19746
mclaren 12 % -P 9500
Traceback (innermost last):
  File "/opt/zope/Zope-2.1.6-src/z2.py", line 493, in ?
  File "/opt/zope/Zope-2.1.6-src/ZServer/medusa/http_server.py", line
552, in __
    self.bind ((ip, port))
  File "/opt/zope/Zope-2.1.6-src/ZServer/medusa/asyncore.py", line 205,
in bind
    return self.socket.bind (addr)
socket.error: (125, 'Address already in use')

[1]    Done                 ./start -P 9500