Fwd: Re: [Mailman-Users] OutgoingRunner Failing

[Brian Greenberg]

David Richards wrote:
> Hi Brian,
> 
> I have seen the same problem as you described in your post, I was running a 
> shared installation via NFS.  With OutgoingRunner's running on each box.
> 
> Would you expect this to have the same results?

While the idea of running MM on multiple machines in itself kind of makes 
my head hurt, I'm pretty sure that you're seeing a similar if not the same 
problem that I encountered.  The logs you've posted look identical.

 From what I could tell from the code, if you are running more instance of 
OutgoingRunner (or any qrunner, for that matter), you *will* have regular 
crashes.  This is because (assuming 4 slices, numbered 0 through 3) each 
slice should manage 1/4 of the queue hash space.  However, as coded, slice 
0 will grab files from the *entire* queue, not just the first quarter. 
This results in a race condition.  The qrunner crash is a result of both 
slice 0 and another slice seeing a file in the last 3/4 of the hash space, 
and both beginning to process it -- one will finish and erase the file, the 
other slice will crash.

Try making the following change:

In "mailman-2.1.5/Mailman/Queue/Switchboard.py", change line 167 from

	if not lower or (lower <= long(digest, 16) < upper):

to

	if (lower == upper) or (lower <= long(digest, 16) < upper):

This completely eliminated my problem.


Brian.

>>Date: Tue, 14 Sep 2004 14:25:42 +0900
>>From: Jim Tittsler <jwt at onjapan.net>  
>>Subject: Re: [Mailman-Users] OutgoingRunner Failing  
>>To: David Richards <d.richards at qut.edu.au>
>>Cc: mailman-users at python.org
>>
>>On Sep 14, 2004, at 09:04, David Richards wrote:
>>
>>
>>>I have the OutgoingRunner process failing quite regularly, and this has
>>>resulted in a build up of mail in the qfiles/out directory.  How do I 
>>>find out
>>>what is going on in this process for it to be failing like this?
>>
>>Are there any clues in your logs/error log?  If you are lucky, there 
>>will be a traceback showing why OutgoingRunner is crashing.
>>
>>(Have you configured Mailman to run with multiple OutgoingRunners in 
>>your mm_cfg file?  If so, check for Brian Greenberg's recent problem 
>>report and fix.)

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+      Brian Greenberg       +     University of Manitoba     +
+   grnbrg at cc.umanitoba.ca   +   ACN -- Unix Software Admin   +
+-------------------------------------------------------------+
+ Tasklist and PGP key at http://home.cc.umanitoba.ca/~grnbrg +
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 248 bytes
Desc: OpenPGP digital signature
Url : http://mail.python.org/pipermail/mailman-users/attachments/20040914/86f88e84/attachment.pgp