soley@ontenv.UUCP (Norman S. Soley) (12/12/88)
I've recently noticed a problem with the .ca domain maps. Every map entry generates at least two lines of pathalias output, the first looks like: subdomain.ca hop1!hop2!hop3!machine!%s Which is exactly what is supposed to happen. The second one: .subdomain.ca hop1!hop2!hop3!%s is bad news, hop3 is not an authority for .subdomain.ca, machine is. You should be safe mailing to user@subdomain.ca and only experience problems mailing to user@internal.subdoms.subdomain.ca Tht is unless you are using smail 2.5. Seems the resolution algorithm is a little botched so that user@subdomain.ca will always resolve to hop1!hop2!hop3!user because it matches the .subdomain.ca line in the path file first. Presto! all mail to any .ca site fails. -- Norman Soley - Data Communications Analyst - Ontario Ministry of the Environment UUCP: uunet!attcan!lsuc!ncrcan!ontenv!soley VOICE: +1 416 323 2623 OR: soley@ontenv.UUCP " Stay smart, go cool, be happy, it's the only way to get what you want"
andrew@dgbt.uucp (Andrew Patrick) (12/17/88)
In article <314@ontenv.UUCP> soley@ontenv.UUCP (Norman S. Soley) writes: ... >Th[a]t is unless you are using smail 2.5. Seems the resolution algorithm >is a little botched so that user@subdomain.ca will always resolve to >hop1!hop2!hop3!user because it matches the .subdomain.ca line in the >path file first. Presto! all mail to any .ca site fails. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ The situation is not quite as bad as you suggest. Let's look at a concrete example. Suppose you wanted to get mail to 'user@vaxr.uwo.ca' (one of the internal machines at the University of Western Ontario). The site 'julian' is the authority for the '.uwo.ca' domain. From here, the pathalias outputs look like this: |.uwo.ca bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!%s |uwo.ca bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!uwo.ca!%s Now looking at the way Smail 2.5 parses the addresses 'user@vaxr.uwo.ca': |resolve: parse address 'user@vaxr.uwo.ca' = 'user' @ 'vaxr.uwo.ca' (DOMAIN) |getpath: looking for '.vaxr.uwo.ca' |getpath: looking for 'vaxr.uwo.ca' |getpath: looking for '.uwo.ca' |route: 'vaxr.uwo.ca' (uwo.ca) = 'bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!%s' (199) |resolve: parse route 'bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!vaxr.uwo.ca!user' = 'bnr-vpa!utgpu!ai.toronto.edu!julian!vaxr.uwo.ca!user' @ 'bnr-rsc' (UUCP) |resolve 'user@vaxr.uwo.ca' = 'bnr-vpa!utgpu!ai.toronto.edu!julian!vaxr.uwo.ca!user' @ 'bnr-rsc' (UUCP) |bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!vaxr.uwo.ca!user Thus, Smail will route the mail to 'vaxr.uwo.ca' correctly. Thus, all mail to any .ca site does NOT fail. The weirdness comes in when Smail parses the address 'user@uwo.ca': |resolve: parse address 'user@uwo.ca' = 'user' @ 'uwo.ca' (DOMAIN) |getpath: looking for '.uwo.ca' |route: 'uwo.ca' (uwo.ca) = 'bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!%s' (199) |resolve: parse route 'bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!user' = 'bnr-vpa!utgpu!ai.toronto.edu!julian!user' @ 'bnr-rsc' (UUCP) |resolve 'user@uwo.ca' = 'bnr-vpa!utgpu!ai.toronto.edu!julian!user' @ 'bnr-rsc' (UUCP) |bnr-rsc!bnr-vpa!utgpu!ai.toronto.edu!julian!user Instead of sending mail to 'julian!uwo.ca!user' as the pathalias output would suggest, Smail sends it to 'julian!user'. Unless julian = uwo.ca (and this is not declared in the maps), this routing will cause a problem. Thus, Smail will get mail to a machine within a .ca domain, but it may not get mail to the domain gateway. {}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{} Andrew Patrick, Ph.D. Communications Research Centre (613) 990-4675 Dept. of Communications, Canada andrew@crcmar.UUCP ...utzoo!bnr-vpa!bnr-rsc!crcmar!andrew BITNET: andrew@doccrc INTERNET: andrew@dgbt.crc.dnd.ca N.B. We are changing from crcmar.UUCP to dgbt.UUCP, but the maps have yet to be updated. {}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
rayan@ai.toronto.edu (Rayan Zachariassen) (12/17/88)
Like Andrew points out, this is a problem with smail 2.x. The semantics of a pathalias output line like: .foo.ca bar!%s is that "bar" knows how to get to the ".foo.ca" domain and its subdomains. Therefore user@baz.foo.ca should turn into bar!baz.foo.ca!user, and user@foo.ca should become bar!foo.ca!user. Smail may be confusing the semantics in this case with that of foo bar!%s which indeed indicates that bar and foo are the same machine. I don't know what the fix for smail is. 3 or 4 different sites have asked me about this recently, thinking it was a problem in the maps. rayan
andrew@dgbt.uucp (Andrew Patrick) (12/22/88)
In article <88Dec17.114819est.38035@neat.ai.toronto.edu> rayan@ai.toronto.edu (Rayan Zachariassen) writes: >Like Andrew points out, this is a problem with smail 2.x. ... ... >I don't know what the fix for smail is. According to recent discussions in the info-smail mailing list, it is the same old story: "It's fixed in Smail3.1"! Smail3.1 is still in BETA test. {}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{} Andrew Patrick, Ph.D. Communications Research Centre (613) 990-4675 Dept. of Communications, Canada andrew@crcmar.UUCP ...utzoo!bnr-vpa!bnr-rsc!crcmar!andrew BITNET: andrew@doccrc INTERNET: andrew@dgbt.crc.dnd.ca N.B. We are changing from crcmar.UUCP to dgbt.UUCP, but the maps have yet to be updated. {}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
rayan@ai.toronto.edu (Rayan Zachariassen) (12/23/88)
In article <984@dgbt.uucp> andrew@dgbt.UUCP (Andrew Patrick) writes:
# According to recent discussions in the info-smail mailing list, it is
# the same old story: "It's fixed in Smail3.1"! Smail3.1 is still in
# BETA test.
Yes, but...
- It is somewhat misleading to claim "fixed in 3.1", since Smail2.x
and Smail3.x have absolutely nothing in common other than the first
5 letters of their name.
- Smail3.1 had this wrong until someone from Texas (of all places)
complained to me about just the problem being discussed. The
fellow was quite convinced the behaviour he was seeing was correct
(he was, after all, beta testing Smail3.x :-), but went back and checked
and got the Smail people to fix their code.
Nothing is a panacea. Well, almost nothing ;-)
Now, if only Geoff and Henry would ship Real C News so the poor sun-stroke
people don't have to read this...
rayanlyndon@auvax.uucp (Lyndon Nerenberg) (12/27/88)
In article <984@dgbt.uucp> andrew@dgbt.UUCP (Andrew Patrick) writes: > >According to recent discussions in the info-smail mailing list, it is >the same old story: "It's fixed in Smail3.1"! Smail3.1 is still in >BETA test. No, it's still in ALPHA test. Anyway, alot of the problem seems to stem from map entries that point to the top level domain name via an alias or direct connection. E.g., the map entry for .nexus.ca shows: alberta nexus.ca(SOMETHING) ncc nexus.ca(SOMETHING) ncc .nexus.ca(SOMETHING) where SOMETHING is a weight I can't remember at the moment, but is equivelent across these three entries. This will cause pathalias to generate paths like: .nexus.ca ...!ncc!%s nexus.ca ...!alberta!%s assuming alberta is a closer site than ncc as far as you are concerned. When smail gets ahold of an address like lyndon@nexus.ca it generates ...!alberta!lyndon, whereas lyndon@ncc.nexus.ca generates a route of ...!alberta!ncc!ncc.nexus.ca!lyndon. People may agree or disagree with this behaviour, however it seems this can be solved by eliminating the entry for nexus.ca from the map, forcing pathalias to generate fully qualified domain names in the route in all cases. This shouldn't break anything as any site advertising itself as a domain gateway should be able to handle the top level domain name as well as any subdomains. By dropping the nexus.ca entry, the route generated for lyndon@nexus.ca should be ...!alberta!ncc!nexus.ca!lyndon. -- Lyndon Nerenberg Computing Services Athabasca University {alberta, attvcr, ncc}!auvax!lyndon || lyndon@nexus.ca
cks@ziebmef.uucp (Chris Siebenmann) (12/30/88)
In article <984@dgbt.uucp> andrew@dgbt.UUCP (Andrew Patrick) writes:
[it's fixed in smail 3.1.]
Unfortunately, Smail 3.1 is trying to replace sendmail, not smail 2.5
and is of an appropriately large size. I'm planning to stick to smail
2.5 unless smail 3.1 gets a lot smaller.
Does anyone know who the official maintainers of smail 2.5 are?
--
"...in all the history of Earth, there's never been a heaven, never
been a house of gods that was not built on human bones."
Chris Siebenmann uunet!{utgpu!moore,attcan!telly}!ziebmef!cks
cks@ziebmef.UUCP or .....!utgpu!{,ontmoh!,ncrcan!brambo!}cksclewis@ecicrl.UUCP (Chris Lewis) (12/31/88)
In article <1988Dec30.012400.18278@ziebmef.uucp> cks@ziebmef.UUCP (Chris Siebenmann) writes: >In article <984@dgbt.uucp> andrew@dgbt.UUCP (Andrew Patrick) writes: >[it's fixed in smail 3.1.] > Unfortunately, Smail 3.1 is trying to replace sendmail, not smail 2.5 >and is of an appropriately large size. I'm planning to stick to smail >2.5 unless smail 3.1 gets a lot smaller. > Does anyone know who the official maintainers of smail 2.5 are? I was conversing with Chip Salzenburg regarding the Xenix patches to smail 2.5 a long way back, and it appears that smail 2.5 is officially orphan. Chip sent his patches off to what he *thought* were the smail maintainers, and they told him that they had dropped 2.5 in favor of 3.1 development/testing 3.1, so Chip posted his stuff by himself. You might want to contact Chip and ask him for further info. I haven't been following this conversation much, but I did snaffle this from the net back in June - maybe this'll help: |>From: honey@umix.cc.umich.edu (Peter Honeyman) |Newsgroups: news.admin,comp.mail.uucp |Subject: Re: Internet Paths in UUCP Maps -- Can We Stop? |Message-ID: <4116@umix.cc.umich.edu> |Date: 5 Jun 88 13:41:09 GMT |Organization: Center for Information Technology Integration, Univ of Michigan |Lines: 5 |Xref: spectrix news.admin:2225 comp.mail.uucp:1196 |UUCP-Path: {uunet,rutgers}!umix!honey | |i run pathalias output through this script: | |egrep -v '(\.(com|edu|mil|gov|net|org|arpa|[a-z][a-z]) .*!.*!)|(.\.(com|edu|mil|gov|net|org|arpa|[a-z][a-z]) )' | | peter -- Chris Lewis, Markham, Ontario, Canada {uunet!attcan,utgpu,yunexus,utzoo}!lsuc!ecicrl!clewis Ferret Mailing list: ...!lsuc!gate!eci386!ferret-request (or lsuc!gate!eci386!clewis or lsuc!clewis)