The mailing list for listmasters using Sympa

[Fabio Napoleoni <address@concealed>]

Patrick von der Hagen ha scritto:
> Am Donnerstag, den 08.11.2007, 03:05 +0100 schrieb Fabio Napoleoni:
> [...]
>> My question is why those messages are truncated? Is it bouncequeue
>> fault, postfix fault or some other host fault?
> I consider it to be normal that some remote server bounces and truncates
> a message. Even if it did not, you might receive a bounce which can not
> be matched to an address in your subscriber-table. If "A" is subscribed,
> "A" forwards to "B" later and "B" is broken (over quota, deleted, etc.),
> you will receive a bounce that "B" failed, but likely without any way to
> automatically recognize that "A" is the culprit.

Ok, I thought the same, but imho the strange thing is that in all these
truncated messages miss the first part, what i found in sympa bounce
directory (so after executing bouncequeue) is the final part of the
message. Now I'm trying to sniff bounce messages, before they are
processed by postfix and bouncequeue, in order to verify their content
at arrival time.

> To handle those cases you have no real way but activating VERP. Given
> your list-size, it is probably unwise to activate it for all
> subscribers, but if that list is used frequently a small percentage
> should be no problem. Of course it takes longer to identify all broken
> subscribers, but since VERP increases the number of individual messages,
> I'd just start with something as low as 2%, watch it and decide later
> wheter or not your system can handle 5% oder even 10%.

Thank you for your answer but, at this time, i use sympa 5.1.2 (which
not support VERP) because in debian stable, sympa 5.2.3 package has a
bug and fail to upgrade. Soon we'll move the newsletter on another
machine that has installed 5.2.3 version.

-- 
Fabio Napoleoni
address@concealed

****************************************************************
 "Computer Science is no more about computers than astronomy is
 about telescopes"
                                             Edsger W. Dijkstra
****************************************************************