[Mailman-Users] Optimizing mailman deliveries

[Harald Meland]

[Jonathan Bradshaw]

> There are various settings that seem to effect how mailman batches deliveries.
> However, I can't seem to find anything that gives me an overview of the
> "method" to this implementation.

Assuming you by "batching" mean how Mailman partitions the recipients
during delivery of one list message into a number of SMTP dialogues
(and not anything that i.e. has to do with when digests should get
sent out):

The actual batching is done by the "deliver" script, installed in the
"scripts" subdirectory (but distributed in the "mail" subdirectory).
At the top of the script this documentation string is found:

: """Partition a mass delivery into a suitable set of subdeliveries.
: 
: The script takes the following protocol on stdin:
: 
:  line 1: batchnum
:  line 2: sender
:  line 3...n+2: n recipients
:  line n+3: <empty> - delimiting end of recipients
:  line n+4: message content
:  
: The recipients will be distributed into at most batchnum batches, grouping
: together recipients in the same major domain as much as possible."""

> I understand that mailman was designed with the (inefficient) sendmail
> system and it may make sense for that but for those of us using qmail or
> (soon) postfix/vmailer are there guidelines for how I might make sure
> the delivery method is optimized for those environments?

I don't understand -- are you implying that when qmail or postfix are
less efficient in delivering mail when it was injected with multiple
RCPT To:s?  I find that hard to believe :)

However, if you are talking of VERP support (which I believe
e.g. ezmlm has) for weeding out bad subscriber addresses, I think this
is one of the things that are "being worked on" for Mailman.  I
wouldn't expect that stuff to go into the 1.0 series, though.

Or do you by "optimized" mean something entirely different?  If so,
please specify.
-- 
Harald