[Mailman-Users] Making all bounces, even temporary ones, go to list owner ASAP

Mark Sapiro msapiro at value.net
Tue Feb 20 02:46:01 CET 2007

Kelly Jones wrote:

>Many of my customers run tightly-controlled lists where it's important
>to know (immediately, if possible) if email to any of the members is
>bouncing (eg, an ISP mailing its customer base), even if the bounce is
>Is there a way to do this using the bounce processing options in
>Mailman? EG, set the warn threshold really low and have all warnings
>be cc'd to the list owner?

There are a couple of settings. You need to set both
bounce_unrecognized_goes_to_list_owner and
bounce_notify_owner_on_disable to Yes. The former sends all messages
received at the -bounces address which aren't recognized and handled
by bounce processing to the list owner. This includes the vacation

This leaves messages which are recognized as bounces. Of these, bounces
for addresses that aren't list members (possibly a DSN reporting a
final delivery address different from the original) will be discarded
as will 'delayed' notices. The rest will be scored. If you set
bounce_score_threshold to 1.0 or less, the member's delivery will be
disabled on the first bounce and the owner notified along with a copy
of the bounce per bounce_notify_owner_on_disable. Now, the owner will
probably want to re-enable delivery for that member if the owner
thinks the bounce is 'temporary'.

Don't be concerned about 'soft bounces' only scoring 0.5. That's wrong
in the documentation. All bounces scored at all are scored 1.0.

Other caveats. If there are any members with bounce scores currently >=
1.0, even if the info is stale, they will be disabled the next time
cron/disabled runs.

If VERP_PROBES is set to Yes in mm_cfg.py, I think it is possible for
the message to bounce for a 'temporary' reason, and the subsequent
probe to not bounce. In this case, the member won't be disabled and
the owner won't be notified of the original message bounce.

>An obvious hack is to tweak the alias generator to do things like:
>listname-bounces: listname-owner
>but is there a cleaner (and "officially supported") way?

The only problem with the above is you lose the benefits of automated
bounce processing. If you can arrange for your MTA to both pipe the
-bounces mail to the wrapper for bounce processing and deliver it to
-owner, you can have both.

Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan

More information about the Mailman-Users mailing list