john@jetson.UPMA.MD.US (John Owens) (08/24/88)
I've run into an interesting anomaly between pathalias and smail (and possibly other pathalias-using routers). First, let me illustrate with a couple of pathalias runs: % cat one me gateway(1) me him(100) him gateway(100) gateway = domain.org gateway .domain.org % pathalias -l me < one me %s him him!%s gateway gateway!%s .domain.org gateway!%s domain.org gateway!%s % cat two me <gateway>(1) me him(100) him gateway(100) gateway = domain.org gateway .domain.org % pathalias -l me < two me %s him him!%s .domain.org him!gateway!%s gateway gateway!%s domain.org gateway!%s Note that declaring the me -> gateway link as terminal causes the path to .gateway.org to take the higher-cost link, which is exactly what pathalias should do. When smail decides to route to user@domain.org, it checks for routes in the following order: .domain.org domain.org .org org smart-host (if defined to) In the first two cases, it uses route!user; in the remainder, it uses route!domain.org!user. Now an actual example, showing why I think there's a problem. uunet declares uunet <rutgers>(DEDICATED) rutgers says rutgers = rutgers.edu rutgers .rutgers.edu The pathalias output for my site contains .rutgers.edu long!path!to!rutgers!%s rutgers.edu uunet!rutgers!%s Ok so far. Now, when smail sees mail for user@rutgers.edu, it uses long!path!to!rutgers!user, instead of uunet!rutgers!user. I think the lookup process for user@domain.org should first check for an exact match for domain.org, and if found, replace the %s with "user". Otherwise, it should check for domain gateways (*only those beginning with a dot*), starting with .domain.org and working down, and if found, replace the %s with "domain.org!user". In other words, it should only look for domain.org (giving route!user) .domain.org (giving route!domain.org!user) .org (giving route!domain.org!user) smart-host (giving route!domain.org!user) when routing user@domain.org and for sub.domain.org (route!user) .sub.domain.org (route!sub.domain.org!user) .domain.org (route!sub.domain.org!user) .org (route!sub.domain.org!user) smart-host (route!sub.domain.org!user) for user@sub.domain.org. This would cause mail for user@rutgers.edu to take uunet!rutgers!user, but mail for user@foobar.rutgers.edu to take long!path!to!rutgers!foobar.rutgers.edu!user, which is as it should be. I'd appreciate comments on this. What's the state of smail 3.0? What other pathalias-using routers do this? Do any do it the way I suggested above? -- John Owens john@jetson.UPMA.MD.US SMART HOUSE L.P. uunet!jetson!john (old uucp) +1 301 249 6000 john%jetson.uucp@uunet.uu.net (old internet)
honey@umix.cc.umich.edu (Peter Honeyman) (08/26/88)
John Owens writes: >When smail decides to route to user@domain.org, it checks for routes in >the following order: > .domain.org > domain.org > .org > org > smart-host (if defined to) if that's true, it's wrong. the order should be domain.org, .org, smart host. peter
soley@ontenv.UUCP (Norman S. Soley) (08/29/88)
In article <114@jetson.UPMA.MD.US>, john@jetson.UPMA.MD.US (John Owens) writes: | I've run into an interesting anomaly between pathalias and smail (and | possibly other pathalias-using routers). | | When smail [2.5] decides to route to user@domain.org, it checks for routes in | the following order: | .domain.org | domain.org | .org | org | smart-host (if defined to) [After some supporting exapmles to support it he makes the following statement] | In other words, it should only look for | domain.org (giving route!user) | .domain.org (giving route!domain.org!user) | .org (giving route!domain.org!user) | smart-host (giving route!domain.org!user) | when routing user@domain.org and for | sub.domain.org (route!user) | .sub.domain.org (route!sub.domain.org!user) | .domain.org (route!sub.domain.org!user) | .org (route!sub.domain.org!user) | smart-host (route!sub.domain.org!user) | for user@sub.domain.org. | | I'd appreciate comments on this. What's the state of smail 3.0? | What other pathalias-using routers do this? Do any do it the way I | suggested above? This bounced around the info-smail mailing list and one of the smail3.1 developers posted a diagnostic trace which shows that the order is now correct (i.e. in smail3.X). The info-smail list (which should really be called info-smail3 or some such) can be reached at info-smail-request@uts.amdahl.com -- 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