romain@pyrnj.uucp (Romain Kang) (12/29/87)
In article <188@tijc02.UUCP> seb022@tijc02.UUCP (Scott Bemis) writes: | 12/28/87 | | Is e-mail getting bounced around, and not being delivered? Or have I done | something wrong? Below is part of the text I received when trying to send | mail. The error message seems that there were too many mail hops, 17 being the | maximum; however, I specified | | rti!mcnc!gatech!ut-ngp!uniq!uniqvax!liveris | | and | | rti!mcnc!gatech!rutgers!clyde!cbosgd!ihnp4!uniq!uniqvax!liveris The clues are in the Received: headers in the sample message you posted with your article. The path your message took is something like: rti-> mcnc-> gatech-> ut-ngp-> ut-sally-> im4u-> rutgers-> im4u-> ut-sally-> ut-ngp-> ut-sally-> im4u-> rutgers-> im4u-> ut-sally-> ut-ngp-> ut-sally-> im4u-> rutgers-> ERROR Both of the paths you specified pass through sites that perform automatic "least cost" routing. I know that one of these sites is rutgers. The other appears to be ut-ngp. I don't have a map entry for ut-ngp; maybe they've recently dropped off the net. ut-ngp doesn't believe it has a connection to uniq (as your first path would suggest) so it decides to punt and passes the mail to someone else to deliver. Somewhere along the line, your message gets passed to rutgers. rutgers looks at the address and decides (as you did) the best path to uniq is through ut-ngp, and routes through im4u!ut-sally to get there. A pattern emerges... Several iterations later, rutgers looks at your message and realizes something awful has happened and tries to be friendly (hmm, it's not going to get to liveris, so seb022 ought to know, so he can fix things...) Peter Honeyman once had an example of such a mail loop on his office door, with the footnote, "If this weren't so funny, I'd be really upset." | In first failed attempt, I used pathalias to generate the most efficient path. | In the second failed attempt, I created the path "by hand" - looking at the | USENET backbone map. Your first attempt was probably the wiser one. Your second attempt didn't work any worse, but a news link does not necessarily imply a mail connection. However, this is academic for the reasons given above. I believe this means that in the anarchic USENET environment, the only way to be sure you will reach your intended host is to make a direct connection to them. Certainly someone could say, "everyone MUST have the latest maps and MUST use smail, and no one is allowed to forcibly reroute ANY messages passing through his site". This might make it possible for *your* message to get through, but there's no way to enforce such regulations. It's very likely such a fiat would break someone else's mail, too. | Below are the error messages I received back. [ see original article for 60+ lines of headers not quoted here... ] -- "If at first you don't succeed, try, try again. Then give up. There's no use in making a blame fool of yourself." --Mark Twain
fletcher@im4u.UUCP (Fletcher Mattox) (12/30/87)
In article <738@pyrnj.uucp> romain@pyrnj.UUCP (Romain Kang) writes: >In article <188@tijc02.UUCP> seb022@tijc02.UUCP (Scott Bemis) writes: >| rti!mcnc!gatech!ut-ngp!uniq!uniqvax!liveris >The other appears to be ut-ngp. I don't have a map entry for >ut-ngp; maybe they've recently dropped off the net. The link ut-ngp!uniq did not exist; the map was wrong. In fact, ut-ngp no longer exists. It has been replaced by ut-emx. The current maps reflect both of these facts.