[mod_python] mod_python losing variables and functions

Jack Neely jjneely at pams.ncsu.edu
Thu Feb 17 23:24:16 EST 2005

On Thu, Feb 17, 2005 at 06:02:57PM -0500, Graham Dumpleton wrote:
> Jack Neely wrote ..
> > My servier running apache/mod_python is RHEL 3.  python-2.2.3-6.1
> > mod_python-3.0.3-5.ent  httpd-2.0.46-44.ent
> > 
> > What really gets me is that I also maintain Current.
> > http://current.tigris.org  This XMLRPC server is based off of the
> > Current code which, to my knowledge has never hit this problem before.
> You are running an old version of mod_python on the probematic site.

'eh...yeah I know.  RHEL3 is our current production level platform...I'd
rather not have to vary too much if I can get away with it.

> What version of mod_python are you running on the site that works all
> the time?

I keep Current running on an FC3 box (mod_python-3.1.3-5) but I do test
on RHEL3 but probably not enough to hit this bug.  I don't have any
users complaining either.

> Do you have anywhere else on the site any other files called API.py and
> handler.py which might get loaded under a different URL?

There is other mod_python code that gets run.  It doesn't use any files
called API.py or handler.py.  Its not an XMLRPC engine at all.  It
builds kickstarts.  I don't see how they would be conflicting.

I'm half way tempted to force the interperter to exit if I detect this
error condition....  Its been frustrating.

The other interesting part of this is that there are 2 servers behind an
LVS.  They both access the same filespace via AFS and do not have
permission to write to those directories.  This situation only seems to
happen on one of them.  I've checked the software on both machines and
I've got good installs.  


Jack Neely <slack at quackmaster.net>
Realm Linux Administration and Development
PAMS Computer Operations at NC State University
GPG Fingerprint: 1917 5AC1 E828 9337 7AA4  EA6B 213B 765F 3B6A 5B89

More information about the Mod_python mailing list