we've been using ZOPE in combination with the Silva CMS for around four
years now to serve our University's homepage. And everything worked fine
A week ago we switched to a new layout (for corporate reasons) and now
we're experiencing frequent crashes of the Zope servers. Fortunately
enough the reconnect themselves to the ZODB but since this is now
happening around every five minutes, I'm rather worried that this might
permanently damage the ZODB.
I have absolutely no idea how this can happen, as we're using the same
python, libxml2, libxslt and other module versions as with the old
homepage - in fact the new site even runs on the same hardware. We never
experienced any problems like these up until now.
As far as I understood so far, it "requires" some C modules to
successfully cause ZOPE to segfault?
Versions we're using:
and a Kerberos Module
plus the Silva CMS (2.1) on top.
We have four ZOPE servers, each running two ZEO processes and a separate
ZODB. The machines all run RedHat Enterprise Linux 5.4. In front of that
Apache, Squid and Pound take care of the caching.
What we did was to examine the coredump-files with gdb but unfortunately
this didn't prove to be very helpful because either things go wrong
during garbage collection or some ceval stuff. So basically something
trashes certain python-objects at time before.
Do you have *any* hinst in how to track down this problem? Or are there
any known problems with the versions above? The changelogs didn't reveal
any plausible cause for me...
Zope maillist - Zope@zope.org
** No cross posts or HTML encoding! **
(Related lists -