[Mailman-Users] Changing bounce_score_threshold punts subscribers?

[Bryan Fullerton]

I'm using Mailman 2.1.5.

I ran into a situation today where a list owner changed the
bounce_score_threshold from 7 to 3, and 55 subscribers were disabled
when cron/disabled ran. Most of these users haven't seen a bounce in
at least weeks, in some cases years, but they all did have bounce
scores over 3 and under 7.

I'm unsure exactly why this happened, I thought there were things in
place to handle this (removing stale bounce info, bounce probe
emails). From my mail logs it looks like Mailman sent an email to the
subscriber and the list owner at the same time. I don't know what was
in the mail to the subscriber, but the list owner got this:

=====
Subject: 	Bounce action notification
Date: 	Tue, 16 Nov 2004 09:00:18 -0500
From: 	mailman at lists.samurai.com
To: 	trawlers-and-trawlering-owner at lists.samurai.com

This is a Mailman mailing list bounce action notice:

    List:       Trawlers-and-trawlering
    Member:     [elided]
    Action:     Subscription disabled.
    Reason:     Excessive or fatal bounces.
    

The triggering bounce notice is attached below.

Questions? Contact the Mailman site administrator at
mailman at lists.samurai.com.

------------------------------------------------------------------------

[disabled by periodic sweep and cull, no message available]
=====

In the Mailman bounce log for that same address I see:

Nov 16 09:00:18 2004 (89086) trawlers-and-trawlering: [elided]
disabling due to probe bounce received

The last info I have for that address in the bounce logs before this is:

Oct 30 00:07:48 2004 (748) [elided]: trawlers-and-trawlering current
bounce score: 4.0

>From what I can see all of the 55 subscribers disabled today are
digest subscribers, but their digest is being successfully delivered
every day.

So... 

Why wasn't their bounce score determined to be stale and discarded?
bounce_info_stale_after is set to 3 (was set to 7 until last night).

Why was it logged that they were disabled due to a probe bounce when
that wasn't the case? My mail server is pretty quick, but I can't see
it sending and receiving back a probe message in the same second (not
to mention there's no bounce in the mail logs).

If this is a legitimate bug, and not just my misunderstanding of how
things work, let me know and I'll open a bug on SourceForge.

Thanks,
Bryan