Return-Path: Received: from K.GP.CS.CMU.EDU by K.GP.CS.CMU.EDU; 15 Jul 88 20:41:34 EDT To: bovik@K.GP.CS.CMU.EDU Subject: [Paul.Birkel@k]: UUCP and addrmap variations Date: Fri, 15 Jul 88 20:41:31 EDT Message-ID: <9080.585016891@K.GP.CS.CMU.EDU> From: Harry.Bovik@K.GP.CS.CMU.EDU Return-Path: Received: from NL.CS.CMU.EDU by K.GP.CS.CMU.EDU; 12 Jul 88 15:54:59 EDT Date: Tue, 12 Jul 1988 12:03-EDT From: Todd.Kaufmann@NL.CS.CMU.EDU To: pab@k.gp Subject: variations on an addrmap (Re: ARGH! Mailers ...) Message-Id: <584726602/toad@NL.CS.CMU.EDU> Obviously ardent is fucked up--they must have two mailer programs playing volleyball or something. First, you should send mail to postmaster@ardent telling her about this. Looking again, I don't see ardent explicitly in the path; tis.llnl.gov sent it to uunet which sent it to ardent... perhaps a message to all three? Depending on how "good" you feel. Second, try to find a path that doesn't use ardent. Here are some heuristics, given the path auspyr!altnet!altmail!altger!unido!prlb2!vub!arti.!steels%tis.llnl.gov It would seem that trying to avould tis.llnl.gov is a good first step, since that is where problems seem to have started. So, let's find another path to auspyr. Typing "uuhosts auspyr" shows, among other things, the following connections: auspyr necntc(DEMAND+HIGH), sdencore(DEMAND+HIGH), lll-tis(DEMAND), imagen(DIRECT), dlb(DIRECT), sci(DIRECT), altnet(DIRECT), jmr(POLLED), sjsqnt(DIRECT), eridani(POLLED), amdahl(DIRECT) I'll pick one, say amdahl, and find a path there: % addrmap foo@amdahl.uucp amdahl!foo@sun.com Now we replace foo with "auspyr!...!steels", and voila, new mail path. This assumes amdahl talks to auspyr, which is probably the case (some connections are only one-way (i think), but I believe them to be very rare). [for some reason, uuhosts doesn't know about amdahl]. Also, there is the uucp mapper "uunet", (which tis.llnl.gov used, btw), that can take (almost) any uucp address, and which is the default address given by addrmap when addrmap doesn't know: % addrmap user@unknown.uucp user%unknown.uucp@uunet.uu.net but uunet takes a lot of traffic, and if everyone starts using it, will probably shutdown (like seismo did). So, try amdahl!auspyr!altnet!altmail!altger!unido!prlb2!vub!arti.!steels@sun.com and maybe through uunet. Maybe try another path to auspyr. Personally, I've never seen anything like x.y.uucp for a host name; I've only seen names of the form x.uucp. This makes me wonder if ..!vub!arti.!steels is right. The uuhosts entry says vub is the gateway to the VUB LAN (in Brussels). vub is then connected to many machines, 5 of which are named arti{1,2,3,5,6}. I don't know what "arti." will do, but it doesn't seem like the correct form for a node name. I'd try ..!vub!arti!steels, ..!vub!steels, or ..!vub!arti.vub!steels, or ask postmaster or vub!erik (see uuhost entry for vub) as to what the correct way to address to steels is and/or have them forward the message for you. I'd have most faith in sending to vub, via the amdahl@sun.com address, and then asking vub!erik what vub.arti.uucp is supposed to mean. Did I confuse you more? I hope you find the above re-routing heuristics helpful... note this does not generate a shorter path (in terms of what?), but the error ardent complained of was because it had been going around in circles, ardent <--> uunet for some reason. Remember, addrmap generates a path considered to be the most optimum with respect to path TIME, not length or true cost. If any of the links are down, addrmap can't help. This is where YOU must start to poke around the uuhost maps.. If I wasn't clear anywhere above let me know. Good luck, and happy mailing todd Return-Path: Received: from ZOG.CS.CMU.EDU by K.GP.CS.CMU.EDU; 12 Jul 88 12:19:03 EDT Date: Tue, 12 Jul 1988 12:11-EDT From: Tom.Lane@ZOG.CS.CMU.EDU To: Paul.Birkel@K.GP.CS.CMU.EDU Subject: Re: ARGH! Mailers ... Message-Id: <584727118/tgl@ZOG.CS.CMU.EDU> Try sunuk!sunbim!vub!arti!steels@sun.com I got this by grepping thru /../pt/usr/netnews/uumap/* (the Usenet map files) for vub (seems its in Belgium...) and looking at its less-used links. It's too bad that you can't get addrmap to give you alternatives to the path it thinks is cheapest; you might ask Brad White if there is a way to get additional paths automatically. Incidentally, the map claims there are machines arti1,arti2,... at that location; I'm not sure that you have a sufficiently accurate address. tom (PS. you think that's bad, bags wanted to send this reply to ardent!MAILER-DAEMON@uunet.UU.NET (Mail Delivery Subsystem) good thing I looked...) Return-Path: Received: by K.GP.CS.CMU.EDU with BradMail; 12 Jul 88 15:06:27 EDT Date: Tue, 12 Jul 88 14:43:22 -0400 (EDT) From: Bradley White Subject: Re: ARGH! Mailers ... To: Paul.Birkel@K.GP.CS.CMU.EDU Message-ID: <584736202.00566/bww@k.gp.cs.cmu.edu> In-Reply-To: Your message of 12-Jul-88 10:11 (general bboard) Paul, > I have *no* idea how to reroute this to a shorter path. Clearly it isn't the length of the path that is killing you - the mail is in a loop: . -> ardent.UUCP -> uunet.UU.NET -> ardent.ardent.com -> ardent.UUCP -> . The link from TIS.LLNL.GOV to auspyr is marked as "DEAD" in the UUCP maps, but addrmap uses it anyway because it doesn't have a better alternative. And then TIS.LLNL.GOV sends it on to uunet, and the loop begins. > HELP? Try . I'm not sure why addrmap is missing the uunet!unido link, but I will check it out. BTW, the only hosts listed in the UUCP maps are arti[12356], so you may also expect some trouble there. Brad Return-Path: Received: by K.GP.CS.CMU.EDU with BradMail; 13 Jul 88 16:26:07 EDT Date: Wed, 13 Jul 88 16:19:56 -0400 (EDT) From: Bradley White Subject: Re: Mailer details To: Paul.Birkel@K.GP.CS.CMU.EDU Message-ID: <584828396.27965/bww@k.gp.cs.cmu.edu> In-Reply-To: Your message of Wed, 13 Jul 88 15:02:34 EDT <26520.584823754@K.GP.CS.CMU.EDU> G'day, > Is there any documentation/heuristics available describing > how to take advantage of the uucp connections database when > all else fails? The best I can suggest is using "uuhosts" or grepping through the raw map files (/../pt/usr/netnews/uumap/[du].*), starting with the host you want to get to and exploring different paths until you find something you know how to reach. Of course, "addrmap" is supposed to make this sort of thing unnecessary, so if you ever really feel the need to do it I'd like to know why "addrmap" was inadequate. > Also, is there any way to get addrmap to offer alternative > paths than its "best" choice? Unfortunately not (ie. short of regenerating the database). Brad