lee@west44.UUCP (Lee McLoughlin) (10/05/84)
This behaviour seems common to all the uucp's I've come across from V7 to S5. For whatever reason if uucp (or more exactly uuxqt) has to reject a mail message rather than aim the rejection at the user who originated it it will be sent to whoever sent it from the immediate neighbour. Normally this will be uucp doing mail relaying. So failed mail aimed at x!y!z!user which gets stomped on at z will (if uuxqt has its evil way) be mailed to y!uucp Gosh! How usefull (I hear you cry)!!!! I was just thought I'd pass on this usefull info'. You can now all rush away and check your uucp mailboxes and work out what to do with any junked mail sent by far away sites aimed at even further away sites. Incidentally I patched this up in UKUUCP but for some reason it doesn't always work. If anyone has a *working* fix...... -- -------------- Lee McLoughlin <UK>!ukc!lmcl, west44!lee UKUUCP support.
honey@down.FUN (10/06/84)
the "working" fix is non-trivial, but here goes (in brief). hack your mailers to give uux a flag and a user name; most common is -a return_address. then have uux spit out a new job control card with that user name on it; most common is R return_address. now hack uuxqt to get the return address user name off the R card instead of the U card. eh voila! if everybody does this, rejected mail will get back to the right luser. DO NOT give in to the temptation of putting the correct return address in the user field of the U card -- i tried this and broke hundreds of uuxqt's nation-wide (see my earlier notes in this forum). peter
lauren@vortex.UUCP (Lauren Weinstein) (10/08/84)
This behavior has been fixed in several ways in many situations. Sites running advanced mailers (sendmail, etc.) usually are smart enough to route mail failure notifications back to the originator to the extent that he/she can be determined in a complex internet environment. A number of newer uucp's (including all of mine) use the new X. file "R" line (return address) convention to determine routing for failed mail, which helps quite a lot and allows multiple hop failed mail routing. --Lauren--
ber@enea.UUCP (Bjorn Eriksen) (10/09/84)
Even if smart mailers as sendmail in 4.2 can handle mail failures, sites should forward (easily done in 4.2BSD) all mail to uucp, news, notes, usenet, postmaster, ... to some responsible system administrator. -- Bjorn Eriksen ENEA DATA Sweden UUCP: {decvax,philabs}!mcvax!enea!ber ARPA: decvax!mcvax!enea!ber@Berkeley
henry@utzoo.UUCP (Henry Spencer) (10/15/84)
> Even if smart mailers as sendmail in 4.2 can handle mail failures, > sites should forward (easily done in 4.2BSD) all mail to uucp, news, > notes, usenet, postmaster, ... to some responsible system administrator. I assume you mean "all failing mail". The above is actually dead easy to do in *any* Unix. "rmail" here is a shell file, not just a link to mail; here's what it looks like: : '@(#)rmail 1.4 of 24 Sept 84' PATH=/bin:/usr/bin ; export PATH tee /tmp/rm$$ | mail -R $* >/tmp/rmm$$ 2>&1 if test -s /tmp/rmm$$ then ( echo rmail $* ; cat /tmp/rmm$$ ; \ echo ----- ; cat /tmp/rm$$ ; echo ----- ) | mail uucp fi rm -f /tmp/rm$$ /tmp/rmm$$ In other words, if anything goes wrong, mail the complaints and the message to "uucp" so it can be handled. The "-R" option to mail is a local addition, to get around the ugly business of having mail's behavior depend on what name is used to invoke it. If you don't have source, link /bin/mail to /bin/realrmail and change "mail -R" to "realrmail"; we ran that way for quite a while. Something like this really should be standard equipment for *any* site that doesn't have something like sendmail. It's simple and works well. -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry