[Mailman-Users] Throttling output

[Patrick Bogen]

On 6/13/06, Steve Campbell <campbell at cnpapers.com> wrote:
> The failure code is 450 4.7.1. I would probably set the SMTP_MAX_RCPTS =
> "some very small number" also to allow only small amounts at one time.
>
> So, assuming that the milter works properly, and it takes one hundred loops
> for the milter to process the entire retry queue over a given time interval,
> would Mailman be OK with this? As I see it, the first run through the
> milter, 99% of the original email would still not go out, then the second
> run would be 98% of the original, etc. Is this going to cause Mailman to
> give up on some of the emails that are last in line and flag them in any
> way?

I think it should be alright. Mailman's logic says that, if it's a
temporary failure, AND none of the recipients could be delivered, AND
mm_cfg.DELIVERY_RETRY_PERIOD has passed (5 days, in my Defaults.py),
then the message will be discarded. The default DELIVERY_RETRY_WAIT is
1 hour, so that gives you 120 chances. Shorten the retry to, say, 5
minutes, and you get 1440 chances. If you deliver, say, 25 mails per
five minutes (if you can configure it that way), then you get to send
36,000 mails. Overlapping mails might be a problem.

(All told, a better solution is to configure your MTA (even if it's an
MTA local to the mailman machine) to accept them all for delivery, and
then throttle it there. The MTA is better equipped for this, since
Mailman doesn't know about throttling, it's not going to blink if it
thinks it's the right idea to discard the message at some point.)

(P.S., please CC the list on all replies.)

-- 
- Patrick Bogen