this mode of breaking is all too well known.
At one time (before RFC 2000), we tried to get a document out that
specified rational behaviour for mail robots, including a near-absolute
prohibition on the robot sending mail where FROM: pointed to an address
that would elicit an automated response (such as the address of the server
itself).
The effort, like too many other good things, was lost in the mists of time.
Note: Some heavily-safeguarded versions like Listserv's confirm messages
can be OK - but in general, adding "-errors" to that From: line is a Good
Idea nearly 100% of the time.
Harald
--On 16. august 2002 19:35 +0000 Bob Braden <braden@ISI.EDU> wrote:
> What happens, apparently, is that a spammer sends a message to the RFC
> Editor server, purporting to come from a site that is in fact another
> automated SMTP servers. Each server is very polite, responding with an
> error message including the entire request message. So they go into a
> loop, being very polite to each other... This loop continues until
> either our disk gets full from the log files or real requests get
> completely squeezed out by this bogus traffic. We do notice, after a
> few days...