[Mailman-Users] archRunner hogging CPU

[Anne Ramey]

Mark Sapiro wrote:
> Anne Ramey wrote:
>
>   
>> Continued digging lead me to FAQ 4.41....how recent is this.
>>     
>
>
> Not very. The FAQ article is about 3 years old, and the list archive
> thread it refers to regards a Mailman 2.0.11 installation
>
>
>   
>> Does 
>> anyone else still run into this on the newer versions?  I'll talk to 
>> some of my list owners and I can change the default, but I was wondering 
>> if I was chasing a false lead?  (Another note, most of the lists are 
>> fewer than 100 members)
>>     
>
>
> You could try killing ArchRunner. If you 'kill -TERM' it, mailmanctl
> won't restart it.
>
> Then if things don't clear out, check for stale locks from the
> ArchRunner process. See
> <http://www.python.org/cgi-bin/faqw-mm.py?req=show&file=faq04.076.htp>.
>
> That chould get things goung normally, but the messages to be archived
> will start piling up in the qfiles/archive queue.
>
> Then you need to figure out what's wrong. Can you pinpoint a specific
> list? If so, you could just try rebuilding its archive with
>
> bin/arch --wipe <listname>
>
> and then restart ArchRunner with
>
> /usr/local/mailman/bin/qrunner --runner=ArchRunner:0:1 -s
>
> or by stopping and starting mailmanctl (IIRC restart won't restart
> ArchRunner if it was SIGTERMed).
>
> Note that rebuilding the archive with bin/arch is not a step to be
> taken lightly as it MAY renumber messages and invalidate saved URLs,
> but if the issue is a corrupt archives/private/<listname>/database/*
> file, there may be no choice.
>
> You may also wish to check the
> archives/private/<listname>.mbox/<listname>.mbox file with
> bin/cleanarch before running bin/arch.
>   
I did both, just to be safe.  The two main offenders now have weekly 
archives and I used the above to rebuild the archives on the most high 
traffic list.  Now the archiving qrunner is caught up, my load is back 
down and everything is great.  Thanks so much,

Anne