[rescue] Re: [geeks] ADMINISTRIVIA: Changes to mail delivery policies

Bill Bradford mrbill at mrbill.net
Wed Mar 13 03:21:57 CST 2002


On Wed, Mar 13, 2002 at 04:12:28AM -0500, Greg A. Woods wrote:
> Why do you spend any time at all dealing with bounces from subscribers
> who have gone unreachable?  See below for what should happen to them....

Because the list management software isnt *perfect* with dealing with
bounces.  ALso, I'm talking both about handing off mail to a MX thats
broken (accepts the mail, but sends me DSNs over and over and over and 
over), then dealing with the DSNs, as well as the ones that my machine 
simply cant reach at *all*.

> If that's what you're worried about then either you've adjusted the
> wrong parameter, or I've assumed you've adjusted the wrong parameter.

No, I'm talking about what I have to deal with that *caused* me to adjust
things from the default of four days.  Its a "vicious cycle" thing with
people and the damn broken Exchange mail servers.  

> If all the MXers are unreachable then the message is in your queue and
> no DSN of any sort is sent anywhere until the five days has expired, at
> which time your MLM software gets one single copy -- not you, but your
> MLM.  If I'm not mistaken most modern MLM packages have the ability to
> automatically unsubscribe or suspend any address that gets more than a
> given number of bounces.  The human list manager should never even peek
> at these messages unless a manual complaint is sent about someone being
> unsubscribed accidentally.

The problem is really the MTAs on the other end that take the mail, but then
have problems delivering it to the end user (*cough* exchange *cough), but
keep spamming me with DSNs while they try to do it.  Mailman doesent 
recognize the DSNs, only bounces.  I end up having to see the DSNs, then
unsubscribe the person before my end keeps sending mail, which will just
time out and not be delivered anyway.

> Adjusting your own MTA's queue retry time limit will not defer one
> single DSN that comes from any other queue.

Kicking it down to 24 hours for the past few days *has* helped "reboot"
the list manager's bounce detection, and gotten rid of some consistently
bad addresses.  I may take it back up to 2 days, then 4, gradually, as I
can prune out "bad apples" from the subscription list that havent been "flaky"
enough to trigger an auto-unsubscribe up to this point.

Like I said, I'm anxiously awaiting MM2.1 and its improved bounce detection
(and lots of other nice features).  I'm also awaiting moving the sunmanagers
list off to someone else later this summer, which will render the whole point
moot.  Its been nice, but the only thing I get "out of it" is anywhere from
500-950 megabytes/day of used outgoing bandwidth.

Sometimes i"m tempted to just auto-bounce all DSNs back to the broken
Exchange servers they came from. 8-)  (YES I KNOW THIS WOULD BE BAD...)

Bill

-- 
Bill Bradford
mrbill at mrbill.net
Austin, TX



More information about the rescue mailing list