pst@comdesign.uucp (Paul Traina) (07/21/88)
I may be wrong about this, but I was under the impression that any
gateway into a domain is responsible for being able to connect to (or at
least route to) any other host in that domain.
How does one take (the simplest example) mail from:
foo.la.ca.us to bar.la.ca.us
Assuming foo does not know about bar? Do we then pass the mail back
up to the "ca" router and it finds & computes a new path to bar?
--
Paul Traina - {uunet|pyramid}!comdesign!pst - comdesign!pst@pyramid.com
To believe that what is true for you in your private heart is true for all
men, that is genius.wisner@killer.DALLAS.TX.US (Bill "Spam Eggs Sausage and Spam" Wisner) (07/21/88)
The US domain has no gateways. killer.dallas.tx.us is really more of a long
hostname than a domain. We do get the advantages of an Internet MX forwarder
(or five), though.
foo.la.ca.us would send mail to bar.la.ca.us just like foo would send mail to
bar. If both are UUCP hosts, which is likely, than it can be assumed
that pathalias routing will do the trick. Internet sites running decent,
newer mailers will have no problem getting mail to sites in the US domain.
For example, killer's primary MX forwarder is ames.arc.nasa.gov; a good
version of sendmail would find that MX record and send mail for killer
through ames. ames does the rest.
--
Bill Wisner
wisner@killer.dallas.tx.us
..!{ames,decwrl,mit-eddie,osu-cis,rutgers}!killer!wisnermark@clout.Jhereg.MN.ORG (Mark H. Colburn) (07/22/88)
In article <397@comdesign.UUCP> pst@comdesign.uucp (Paul Traina) writes: >I may be wrong about this, but I was under the impression that any >gateway into a domain is responsible for being able to connect to (or at >least route to) any other host in that domain. > >How does one take (the simplest example) mail from: > foo.la.ca.us to bar.la.ca.us > >Assuming foo does not know about bar? Do we then pass the mail back >up to the "ca" router and it finds & computes a new path to bar? In this particular case, foo does not need to know about bar, but the domain naming conventions state that the site listed to the right of a dot in a domain style adress knows how to get the the site immediately to the left of the dot. (This is a bit of a generalization, but it will work for the current discussion...) Therefore '.us' can get to '.ca', '.ca' can get to '.la' and '.la' can get to 'bar'. In your example, '.la' also knows how to get to 'foo'. Therefore, the mail should only have to go up to the server for '.la.ca.us' and then back down to 'bar' if 'foo' and 'bar' do not have a link. Given the following example, foo.sf.ca.us to bar.la.ca.us The mail would have to take the following route (assuming, of course, that foo and bar do not have a direct connection, and that bar and sf.ca.us don't have a direct connection): foo.sf.ca.us -> sf.ca.us -> ca.us -> la.ca.us -> bar.la.ca.us This is the beauty of hierarchical namespace, connections to lower level domains are gaurenteed (well, assuming machines are up :-). The difficulties come in when you branch across top level domains (i.e. mail from foo.la.ca.us to jhereg.mn.org). In this case, all top level servers know how to get to all other top level servers. So in the example above, us knows how to get mail to org, who then passes it down to mn.org and finally to jhereg.mn.org. -- Mark H. Colburn mark@jhereg.mn.org, ..!chinet!jhereg!mark They didn't understand a different kind of smack was needed, than the back of a hand, something was always needed.
wcf@psuhcx.psu.edu (William C. Fenner) (07/22/88)
In article <309@clout.Jhereg.MN.ORG> mark@clout.Jhereg.MN.ORG (Mark H. Colburn) writes: |In article <397@comdesign.UUCP> pst@comdesign.uucp (Paul Traina) writes: |>I may be wrong about this, but I was under the impression that any |>gateway into a domain is responsible for being able to connect to (or at |>least route to) any other host in that domain. |> |>How does one take (the simplest example) mail from: |> foo.la.ca.us to bar.la.ca.us |> |>Assuming foo does not know about bar? Do we then pass the mail back |>up to the "ca" router and it finds & computes a new path to bar? | |Given the following example, | | foo.sf.ca.us to bar.la.ca.us | |The mail would have to take the following route (assuming, of course, that |foo and bar do not have a direct connection, and that bar and sf.ca.us |don't have a direct connection): | | foo.sf.ca.us -> sf.ca.us -> ca.us -> la.ca.us -> bar.la.ca.us | |This is the beauty of hierarchical namespace, connections to lower |level domains are gaurenteed (well, assuming machines are up :-). The |difficulties come in when you branch across top level domains (i.e. mail |from foo.la.ca.us to jhereg.mn.org). In this case, all top level servers |know how to get to all other top level servers. So in the example |above, us knows how to get mail to org, who then passes it down to |mn.org and finally to jhereg.mn.org. | But the problem is, who is going to be sf.ca.us, ca.us, and la.ca.us? I would think that most of the machines in .us would be personal machines, because those are the ones that mostly don't fit. Let's say I decided to register my machine under .us. I would be hogbbs.SCL.PA.US or something like that. Now, let's say I'm the only machine in PA under .US, let alone State College. So who do I pass to? I have to be scl.pa.us as well as pa.us. And is there (going to be?) a US machine that is at the top of the .US domain? Bill -- Bitnet: wcf@psuhcx.bitnet Bill Fenner | Internet: wcf@hcx.psu.edu | This space UUCP: {gatech,rutgers}!psuvax1!psuhcx!wcf | for rent Fido: Sysop at 263/42 |
matt@oddjob.UChicago.EDU (Ka Kahula) (07/25/88)
I see some misinformation floating around news.sysadmin which I want to correct. If a host with a name like w.x.y.z exists, there need not be any hosts with the names x.y.z, y.z, or z. There will generally be a name server FOR the domain z, and possibly some subdomains of z, but those name servers need not be within the domain z. Examples: tycho.yerkes.uchicago.edu exists and is the primary name server for domain yerkes.uchicago.edu. However, there is NO HOST with the name yerkes.uchicago.edu, nor uchicago.edu, nor edu. Name servers for these DOMAINS do exist, but the names of the servers are not identical to the names of the domains. Host clout.chi.il.us exists, but there are neither hosts called chi.il.us and il.us nor name servers for such domains. Name servers for us exist, but they are not themselves in domain us. Those servers COULD choose to give away authority for some subdomain(s) of us, but they do not, in fact, do so. When a random internet host a.b.c.edu (which uses the domain lookup system) wants to send mail to clout.chi.il.us, it does not pass the message up and down some ladder of hosts, but rather it tries to find the correct "mail exchanger" (MX). If a.b.c.edu does not already have any information cached about the us domain, it will ask a "root" server "Who are the MXs for clout.chi.il.us". It will not get a direct answer but will be told "The NSs (name servers) for us are venera.isi.edu and vaxa.isi.edu". a.b.c.edu will then repeat its query to one of those servers, getting the reply, "There is one MX for clout.chi.il.us with preference level of 10 and name oddjob.uchicago.edu". a.b.c.edu will then attempt to connect to oddjob with SMTP and pass the mail message. In general, it will not try any other method to deliver the mail. Mail routing AMONG uucp-only hosts in the us domain is a different question, and one which must be settled by the uucp managers themselves. It will probably be decided in different ways in different places, but it will not be simply "up the ladder/down the ladder" because sites like "il.us" do not exist. ________________________________________________________ Matt Crawford matt@oddjob.uchicago.edu