[Mailman-Users] Bounces threshold never reached

Kala Balik kalabalik at gmx.net
Tue Jun 7 03:19:42 EDT 2016



Am 07.06.2016 um 05:01 schrieb Mark Sapiro:
> On 06/06/2016 01:48 AM, Kala Balik wrote:
>>
>> two weeks ago we sent out a mailing through our six new mailman lists
>> with roughly 6.000 subscribers. We expected to get a number of bounces,
>> however, while the bounces log shows that about 600 subscribers did get
>> a bounce score of 1.0 or 2.0, not a single address reached the threshold
>> of 5.0 (or anything above 2.0) and, consequently, the admins did not get
>> notice about any disables or removals.
> 
> 
> That's exactly what I would expect.
> 
> 
>> Here is my bouncing configuration:
>>
>> bounce_processing: yes
>> bounce_score_threshold: 5.0
>> bounce_info_stale_after: 7
>> bounce_you_are_disabled_warnings: 3
>> bounce_you_are_disabled_warnings_interval: 7
>> bounce_unrecognized_goes_to_list_owner: yes
>> bounce_notify_owner_on_disable: yes
>> bounce_notify_owner_on_removal: yes
> 
> 
> Starting with Mailman 2.1.19, there is also a
> bounce_notify_owner_on_bounce_increment  setting which will notify the
> owner any time the score is incremented.
> 
> 
>> >From reading the information in the web interface my understanding of
>> the bounces processing is that for "bad" (i.e. unreachable) addresses
>> the bounces score increases EVERYDAY by 1.0 (hard bounces) or .5 (soft
>> bounces).
> 
> No.
> 
> The score is incremented by 1.0 on the first received bounce of the day.
> On a day that there are no posts and hence no bounces, the score isn't
> changed.
> 
> If there are no bounces for bounce_info_stale_after (in your case 7)
> days, the next bounce will reset the score to 1.0.
> 
> There aren't actually any 'soft' bounces so the score never increments
> by 0.5.
> 
> 
> Since my bounces log only displays whole numbers (either 1.0
>> or 2.0) I assume that most of my bounces were hard bounces (scoring 1.0
>> each). Therefore, constantly unreachable addresses should have reached
>> the threshold of 5.0 after 5 days.
> 
> 
> After 5 days on which new bounces were recorded. See above.
> 
> 
>> Yet my log reads as if the email providers of most (about 500)
>> subscribers with bounces were unreachable on the day of the mailing
>> (kind of plausible), while the others (about 100) were "down" for two
>> days and then online again (not very plausible). Providers may be
>> unreliable sometimes, but not that unreliable. And our address base may
>> be good, but not that good (no permanent bouncees at all - or am I
>> misunderstanding the concept of permanence here - and, consequently, no
>> automatic unsubscribes at all).
> 
> 
> Your misunderstanding is that Mailman has no way to determine if a
> bouncing address will bounce the next day without sending a post and
> having it bounce. No post = no bounce for that day = no score increment.

It seems that my real misunderstanding is even more fundamental. I was
expecting Mailman to retry sending an email if it received a bounce for
that adress. I thought it would retry until either no new bounce was
received anymore or the bounces threshold was reached. But this really
is a misunderstanding, right? Mailman does not retry sending emails, right?

> 
> 
>> What am I missing here? And how do I "repair" bounce processing and
>> automatic unsubscribe?
> 
> 
> It's working as intended. If yo actually want member's delivery disabled
> after one or two bounces, set  bounce_score_threshold to 1 or 2.
> 
> In general your settings which are the default are appropriate for lists
> which receive several posts a week or at least never more than 7 days
> between posts. If your list is lower volume than that, you may want to
> decrease bounce_score_threshold and increase bounce_info_stale_after.
> 
> Note that once a member's delivery is disabled by bounce_score_threshold
> being reached, Mailman no longer looks for bounces for that member and
> the bounce_you_are_disabled_warnings and
> bounce_you_are_disabled_warnings_interval settings take over.
> 


More information about the Mailman-Users mailing list