[Mailman-Users] Messages just began to be shunted, can't figure out why.

[Mark Sapiro]

Gary Goldberg wrote:

>Hello. I've had a long running Mailman installation with about a dozen lists
>spread over four domain names on a Dell PowerEdge with 3GB RAM running CentOS
>3.9 (fully patched). We're using the non-default Sendmail 8.14.1. I've upgraded
>to 2.1.11 in June and everything was working fine/clean until last Friday
>night.  We had a problem with our SpamAssassin subsystem dropping offline for a
>few hours, and we rebooted the machine to reset everything.


Sometime people make changes (or yum updates) that are not effective
until a reboot or service restart but don't reboot or restart. Then
later, a reboot is done for some reason and no one remembers what was
changed that now causes a problem.


>Since then, messages I attempt to post to a Cub Scout list hosted in Mailman
>have been shunted. Unshunting them just requeues them again. This is the error
>message text I'm getting:
>
>-----------------
>Aug 18 11:19:54 2008 (14837) Uncaught runner exception: recips
>Aug 18 11:19:54 2008 (14837) Traceback (most recent call last):
>   File "/home/digimark/www-mailman/Mailman/Queue/Runner.py", line 120, in _oneloop
>     self._onefile(msg, msgdata)
>   File "/home/digimark/www-mailman/Mailman/Queue/Runner.py", line 191, in _onefile
>     keepqueued = self._dispose(mlist, msg, msgdata)
>   File "/home/digimark/www-mailman/Mailman/Queue/IncomingRunner.py", line 130, in _dispose
>     more = self._dopipeline(mlist, msg, msgdata, pipeline)
>   File "/home/digimark/www-mailman/Mailman/Queue/IncomingRunner.py", line 153, in _dopipeline
>     sys.modules[modname].process(mlist, msg, msgdata)
>   File "/home/digimark/www-mailman/Mailman/Handlers/AvoidDuplicates.py", line 39, in process
>     recips = msgdata['recips']
>KeyError: recips
>
>Aug 18 11:19:54 2008 (14837) SHUNTING: 1219015598.1364529+cd4fcf44c07ffa9b1b0bfa53882233f59121be5f
<snip>
>-----------------


When the message gets to the AvoidDuplicates handler, it has no
recipient list in the message metadata. In the normal pipeline,
AvoidDuplicates is immediately preceded by CalcRecips which ensures
there is a 'recips' list (possibly empty). Apparently, that isn't
happening with this list.

Do other lists work?

The timestamps on the tracebacks (including those I snipped) are all
the same indicating that these are probably from an unshunt. These
messages can never be successfully unshunted because the pipeline in
the metadata of the shunted message starts with AvoidDuplicates, but
the metadata has no 'recips' list, so even after fixing the underlying
problem, you won't be able to simply unshunt these messages. Anyway,
is the traceback from an initial post the same?

So, what is the traceback from an initial post? If it is the same, and
other lists work, do 'bin/dumpdb
lists/<cub_scout_listname>/config.pck' and see if the output has a
'pipeline' attribute and if so, does it have CalcRecips preceding
AvoidDuplicates?

-- 
Mark Sapiro <mark at msapiro.net>        The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan