Re: being dropped from the list
On 17-Sep-98 Hanno Wagner \(Debian-Listmaster-Mails\) wrote:
> On Wed, Sep 16, 1998 at 09:37:04PM -0000, Darren Benham wrote
> On 16-Sep-98 Ralph Winslow wrote:
>
>> It seems that everytime my ISP has the slightest interrruption of
>> e-mail service, I get unsubscribed to this list. Might it be
>
> you should define slightest :-) Our listserver count the
> bounces someone is having at (actually) at the count of 10
> he unsubscribes him from the particular list. debian-user
> is a high-traffic list and the count of 10 is reached
> easily.
I've been watching this thread. Now that you've replied, Hanno, I wish to
add my voice to support the original complaint.
A while ago, your list server wrote to me:
***Your mail address Ted.Harding@nessie.mcc.ac.uk has been removed
***from the debian-user@lists.debian.org mailinglist.
***It generated an excessive amount of bounced mails.
***>From: MAILER-DAEMON@murphy.debian.org
***>To: debian-user-request@lists.debian.org
***>Subject: failure notice
***>
***>Hi. This is the qmail-send program at murphy.debian.org.
***>I'm afraid I wasn't able to deliver your message to the following
***addresses.
***>This is a permanent error; I've given up. Sorry it didn't work out.
***>
***><Ted.Harding@nessie.mcc.ac.uk>:
***>130.88.200.93 does not like recipient.
***>Remote host said: 550 relaying to <Ted.Harding@nessie.mcc.ac.uk>
***>prohibited by administrator
***>Giving up.
On the other hand, nothing seemed to be wrong when I myself looked into
our local mail system (and specifically 130.88.200.93, which DID like me).
Why so?? I asked my local admin friend. To which the reply was:
***This was a temporary problem between 0700 and about 0930 yesterday
***morning. I had a real tantrum about it... but it won't've done any
***good...
So, as a contribution to "you should define slightest :-)" I offer the
above snippet of empirical evidence -- namely a temporary local
misconfiguration lasting a mere 2-1/2 hours can cause people to be
dropped from the mailing list, and have to re-subscribe.
Now I know you have a big mailing list there, which inevitably generates
many bounces because of the amount of traffic. However, the important
parameter in determining whether a user is really unmailable is not the
_number_ of bounces but the length of time, combined with the reason, for
which the remote address generates bounces.
So, therefore, a "retry" bounce generated say every 5 hours usually
indicates that some host on the route is temporarily down, and re-try for
at least 24 hours (and most mailers keep trying for up to 5 days) is in
order. On the other hand, an "unknown-user" bounce is more indicative
of genuine unmailability.
Even so, on one of the lists I run (which has several subscribers in the
FSU, Central Asia and Turkey) it is not uncommon for an address to become
"unknown user" for a few days at a time, and then to rise again from the
ashes when someone out there puts the system right again. So (admittedly
on a list with under 1000 members) my own policy is to check that it
remains "unknown user" for at least 5 days.
Similarly (but for a longer period) for "mailbox full" bounces.
Admittedly such an approach is far more feasible using "bare hands" when
there are few users. But nowadays list-server software can be prgrammed
to handle these things more automatically, and the above-mentioned
mailing list (which now runs under L-Soft LISTSERV) automatically probes
suspect addresses for a while, repeatedly, without involving the list
itself, and automatically removes persistently recalcitrant adresses,
where "persistently" is of course configurable and could be anything from
a few hours to a few weeks.
What is finally needed for a smooth integration with the list itself is
for the list server to set suspect addresses to "no-mail" or equivalent on
the list itself, so that the list does not generate bounces; meanwhile the
list server probes in the background. If the address comes back to life,
then the address is restored to "mail" on the list and all is well.
Otherwise, in due course the list server removes the address (or sends a
message to the list-owner requesting approval for removal).
Either way, however, junking users who become unmailable for very short
periods (such as the above) is definitely unreasonable, in my opinion.
I hope this is a helpful contribution to the discussion.
Best wishes,
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding@nessie.mcc.ac.uk>
Date: 17-Sep-98 Time: 12:06:23
--------------------------------------------------------------------
Reply to: