[Mailman-Users] Large files and qrunner hanging with Mailman-2.0...

[Justin Erenkrantz]

Hello all,

We have seen a problem with the Mailman-2.0 series (with 
Python-2.0 on Solaris-x86) with large .msg files (typically over 
5MB) with attachments which cause qrunner to go into an 
infinite loop (I've seen it run over a weekend to no avail).  
Qrunner holds the lock and hence causes everything else to 
stop until I kill off qrunner, clear out the locks, and 
remove the offending message (the db and msg file).

I believe that I have been able to track the problem
down to the HandlerAPI.DeliverToList call (through
a variety of debugging calls and syslog calls).  I don't
have enough time right now to track it down further.

Attaching to the qrunner process via truss after it goes into 
this bad state shows that there is a constant stream of
brk() system calls.  (Guessing that it is doing some sort
of heap allocation?)

Has anyone seen this?  Are there any known fixes?  Any tips
to try and isolate this problem further?  Or, do I need to 
write a script to watch qrunner and kill it off?  (I saw
Barry's post on mailman-developer regarding the rework of
qrunner.)  It's getting to be a real pain-in-the-butt.  
=-)  We can't limit the size of the messages (we tried, but 
we got overruled by management on that one).

We've seen this problem with Mailman-2.0b3, and I upgraded to
2.0 final on Friday - to no avail...

Thanks in advance,
Justin Erenkrantz
jerenkrantz at ebuilt.com