At 2:55 PM -0500 12/20/01, Irwin Lazar is rumored to have typed:
Short of removing all 30-40 list members in the att.com domain, is there any way to trace this below message back to a specific mailbox that can be removed?
Welcome to the need for...THE PROBE. Not a superhero, not a medical procedure, but a vital part of the listmaster's toolbox.
I have no idea how many of us have written some routine or another to "probe" addresses when things go bad and mailers are brain-damaged, or more frequently in my case some subscriber sets up a four-way address relay that breaks somewhere in the middle without complete headers to track it back, or most recently some poor schmoe just got a wireless account and discovered the guy who had the number before had a forwarder set up to deliver a _digested_ list to it (with, of course, no available Received: headers). I usually allow the problem bounces to collect until I have a handfull of 'em before sending a probe, but some folks I know send one out every month whether they need one or not.
I doubt the one I wrote would do you any good (I hacked it out in Frontier on a Macintosh), but there are, I think, prober software scripts freely available out there. Basically, you want to send mail to either a subset, or your entire distribution list, with seperate messages to each address coded in some way (I just place the address I'm sending to in the first line of the body; some sooper-sekret code, eh?) so you know when you get the bounce that is header-useless what address bit the big'un.
There should be pre-written perl probers out there, at the very least, although it would be way cool to see something written in C for eventual distribution with the SmartList package (a temporary replacement for choplist that placed the line number of the address in the prome text, maybe? I dunno, I'm talking out of my hat here). In the example you gave, I'd send a probe to all of the att.com addresses first, since you have likely (but not necessarily definitively) narrowed the problem down to one of their addresses; if you find it, cool, if not, branch out and probe the rest of your list. (You can use the probe to give basic subscribe/unsubscribe information, too, or remind subscribers of posting rules, or anything else you might want to infrequently remind your subscribers about - this way, even on the 99.9% of good addresses, the probe still has some value.)
Charlie