[Mailman-Users] Runaway python processes

[John Jones]

Gentlemen,

We are noticing a recurring (every few days this happens) number of runaway
Python threads on our mailing list server, and was wondering if there is any
known reasons why this would be happening or if anyone has encountered this
before?  The threads are chewing up CPU time and I just killed 4 of them,
each in the hundreds of minutes of CPU time, the load average on the server
was skyhigh and crawling.

Now correct me if I am wrong, but I think that the queue still fills when
this is happening, which was creating a logjam.  qmail-remote processes are
still spawning, so things are running again, but while I am right now
reading through the documentation and trying to understand the code (not an
accomplished programmer), I would appreciate some help here.

Would creating a cron job that scans for runaway Python processes (running
for more than 10:00, for instance) be okay, or does that break something?

Mailman:  2.0.5
Python:   Active Python 2.1, build 210
Linux:    Redhat 6.2 (Zoot)



John Jones, Systems Admin
http://www.ActiveState.com
Programming for the People