[Mailman-Users] A quick question about slicing outgoing qrunner

[Chris Nulk]

Hi Mark,

Thanks for the information.  I plan on looking at the MTA. 
Unfortunately, when we migrated Mailman to a new server, I don't have 
the same abilities to effect changes as I did on the old server.  The 
server is now managed by a different group and I just run an 
"application" on their server.  Makes identifying and fixing problems a 
major pain in ...

No problem setting the number of queues to a power of two.  I used 3 as 
an example.  Thanks for you input, Barry, for clarifying it.

Jim, the settings you are using look interesting.  I will have to think 
on it.

The biggest issue I have right now is modifications to the MTA.  We are 
using sendmail (it was used on the old server).     When we migrated 
Mailman to the new server, I had to help them setup sendmail since I set 
it up on the old server.  The admins don't know sendmail or postfix 
except on a very minimal level.  It is not part of their normal 
responsibilities.

Anyone willing to take this off-list and help me figure out how to 
switch from sendmail to postfix?  Every time I have tried postfix, it 
has ended very miserably. :)

Thanks,
Chris

On 9/7/2016 4:23 PM, Mark Sapiro wrote:
> On 09/07/2016 02:54 PM, Chris Nulk wrote:
>> I have read the archives regarding slicing the qrunner queues. Mailman
>> here runs on a single system (virtual) along with the local MTA and web
>> server.  Our outgoing queue recently got bogged down.  I would like to
>> increase the number of queues for the outgoing qrunner to three instead
>> of one.
>
> The first thing you need to understand is why the outgoing queue got
> "bogged down". Depending on the actual cause, increasing the number of
> OutgoingRunner slices may not help. Usually, doing things in the MTA is
> more effective. The most effective thing often is setting up a separate
> submission port for Mailman that bypasses  lot of checks.
>
>
>> My question is will I encounter any problems with changing (in
>> mm_cfg.py) the QRUNNERS entry to:
>>
>> QRUNNERS = [
>>      ('ArchRunner',     1), # messages for the archiver
>>      ('BounceRunner',   1), # for processing the qfile/bounces directory
>>      ('CommandRunner',  1), # commands and bounces from the outside world
>>      ('IncomingRunner', 1), # posts from the outside world
>>      ('NewsRunner',     1), # outgoing messages to the nntpd
>>      ('OutgoingRunner', 3), # outgoing messages to the smtpd (change to
>> three '3' queues)
>>      ('VirginRunner',   1), # internally crafted (virgin birth) messages
>>      ('RetryRunner',    1), # retry temporarily failed deliveries
>>      ]
>
> That should be fine assuming you haven't set
>
> USE_MAILDIR = No
>
> but see caveat. Another way which doesn't depend on USE_MAILDIR or a
> future release not adding or deleting a new queue is
>
> QRUNNERS.remove(('OutgoingRunner', 1))
> QRUNNERS.append(('OutgoingRunner', 3))
>
> The QRUNNERS list is not order sensitive so you don't need to insert the
> new entry where the old one was removed.
>
> Caveat: Documentation says the number of slices must be a power of two.
> This is based on the theory that slice membership is based on the last
> bits of the queue entry name (hash), but it hasn't been that way for
> ages if ever. I'm certain the code works with 3 slices, but that's
> relatively untested.
>