[Mailman-Users] inactivating bounce processing to speed up list performance

[Ralf Hildebrandt]

* Lluis Montoliu <montoliu at cnb.csic.es>:
> Thanks Ralf, I understand.
> 
> Apologies If I did not make myself understandable. I am not sending
> garbage to non-existing addresses. All addresses are verified and
> existing but not all messages appear to be delivered equally well, to
> my surprise, while some servers appear to refuse messages coming from
> this list for a while, until they are accepted for distribution, thus
> delaying overall mailman performance while it keeps trying to deliver,

Mailman Performance is not depending on this. Mailman hands the
messages down to an MTA which does the sending - and MTA and Mailman
are not really tightly coupled. Any decent MTA will back off from
temporarily unreachable destinations. If you must, you can configure
your MTA to pass these "not immediately undeliverable mails" to a
fallback relay (config parameter fallback_relay = hostname in Postfix).

> therefore I was considering to cancel bounce processing to speed up
> mailman performance. 

But what you're describing is not generating bounces! Those tempfails
only delay outgoing mail. No bounces occur (unless the delay exceeds a
VERY long time, like 3-5 days - then the mails will eventually bounce).

> When I ask my hosting provider for the SMTP logs there appears to be
> activity by mailman, bounce processing, when there are no messages to
> distribute through the list, therefore I inferred that we are
> experiencing delays in distribution, beyond 30-60 minutes, from a
> message being sent to a message being distributed, because of this
> bounce processing actions. This is a list with about 600 members.

This is really a tiny list. Are you really getting this many bounces?
How many are you getting, and WHY?

I think one needs to look at these logs. 

Bounce processing ONLY kicks in if the mails are REALLY undeliverable
(user unknown, blacklistings and the like). And they would only hurt
performance if a really large percentage of mails would bounce
simultaneously.

Even with 600 members on the list, and all bouncing, processing time
would be REALLY low:

Nov 19 12:24:14 albatross postfix/local[26944]: 3Slv3L0yJvzPbP: to=<mailman-bounces at mail.python.org>,
orig_to=<mailman-bounces at python.org>, relay=local, delay=0.08, delays=0/0/0/0.07, dsn=2.0.0, status=sent (delivered to command:
/usr/local/mailman/mail/mailman bounces mailman)

a single bounce takes 0.08s, with 600 recipients all bouncing you
would be at 600*0.08s=48s - which is not really that much.

-- 
Ralf Hildebrandt
  Geschäftsbereich IT | Abteilung Netzwerk
  Charité - Universitätsmedizin Berlin
  Campus Benjamin Franklin
  Hindenburgdamm 30 | D-12203 Berlin
  Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962
  ralf.hildebrandt at charite.de | http://www.charite.de