[ih] history of bounce processing, was update about bogus list unsubcribe requests

[Joe Touch]

Hi, John,


On 8/26/2016 1:55 PM, John R. Levine wrote:
>>> If this list is like many other old mailman lists, it does no bounce
>>> pruning at all, with predictably unfortunate results.
>> We do use bounce processing, FWIW. It simply failed to work for this
>> address, for whatever reason.
>
> OK.  Having looked at the bounce code in some list managers, I'm not
> surprised.
>
> So since we're talking about Internet history anyway, ...
>
> I assume that back in the heroic era when lists were just large MTA
> aliases, the bounces went back to the human manager who did whatever
> was needed.  Since most of the bounces were actually those obnoxious
> sendmail "It's been two hours and I'm still working on it, and will
> tell you again two hours from now", the usual action was nothing.
I can configure the system so I get a copy of the bounce messages, but I
get a very large number of these for too many reasons -  and many aren't
useful.

E.g., the list has many addresses that are themselves lists, and I
sometimes get bounces when delivery to those members fail. There often
isn't a trail to indicate which subscriber caused the problem (i.e.,
that's where "Apparently-To:" would have helped).

Because of that and the number of lists I run means that the result is
too much noise to see a signal for this sort of exceptional case.

Joe