[Mailman-Developers] thoughts on bounce processing

[James Ralston]

On 2006-05-11 at 14:34-05 Brad Knowles <brad at stop.mail-abuse.org> wrote:

> At 1:24 PM -0400 2006-05-11, James Ralston wrote:
> 
> > The more I think about this, the more I think that silently
> > discarding bounces is an error.
> 
> Mailman doesn't silently discard bounces.

It does if bounce_processing is set to "no".  That's my point: turning
off bounce processing should mean "forward the bounces to the list
administrator(s)", not "discard bounces".

> > There should be two (and only two) options:
> > 
> > 1.  Mailman processes all bounces.
> > 2.  All bounces are forwarded to the list administrator(s).
> 
> In your example above, that would only be appropriate if the AA is
> the list administrator.  I think a better method would be to have
> all bounces sent to the poster, regardless of who that poster is.
> If it's an announcement-only list, then presumably you have a
> restricted set of people who can post to that list, and any one of
> them should be able to handle bounces sent directly back to them.

An intriguing thought.  But...

> This would imply that the envelope sender is left unmodified by
> Mailman, and that could potentially cause problems with things like
> SPF or DKIM, which the sender would need to be aware of.

If the envelope sender is unmodified, then Mailman can't use VERP,
which means that the likelihood that the bounce will contain useful
information (such as, which address actually bounced) is reduced.  :(

I suppose Mailman could stuff the original envelope sender header into
another header (e.g., "X-Mailman-Original-Env-Sender") and then pull
it back out when processing a bounce, but that smells rather hackish
to me.

Perhaps VERP could be enhanced to encode both the original envelope
sender and the recipient?

> > If list admins *really* want to (effectively) discard all bounces,
> > they can e.g. set bounce_info_stale_after to 1 and then set
> > bounce_score_threshold to something impossibly high
> > (e.g. 500,000).
> 
> Keep in mind that I've seen mail loops quickly spin into the
> thousands in a matter of minutes, especially when the messages are
> passing through systems that scrub what they consider to be
> non-useful headers like "Received:".
> 
> I don't know that there's much of anything you could do about such
> loops.  I think the only thing I'd say here is that I'd tend to be a
> little more conservative in terms of what I'd call "impossibly
> high".  ;)

Point taken.

But situations like this are what the "emergency moderation" setting
is for, yes?

James