phil@amdcad.UUCP (Phil Ngai) (01/10/87)
Ok, there has been some criticism of my use of decwrl as an example of
how smail doesn't work even with a well run site. I still think decwrl
is a good example despite Mark's claims that they are an ENET site
first, ARPA second, and only third for UUCP. It is true that decwrl is
all three of those and probably more. But decwrl is (as far as I know)
THE corporate mail gateway. I see it as saying the front wheels on my
car come first and if my back wheels malfunction, well they're not as
important. I don't think that's how DEC does it. And any other
problems I have brought to the attention of the decwrl staff have been
resolved. Except for this one.
So you don't like decwrl. How about ucbvax? I even looked at its
sendmail.cf. It has a special case for amdcad. But it still likes to
take that extra hop through sun. Am I to believe that ucbvax is poorly
run too? If neither DEC nor Berkeley can get it right, what about the
rest of the world? I can't believe anyone would say "oh that extra hop
over the ARPAnet doesn't matter." It seems like a very poor design to
me.
I would like to understand the nature of this problem a bit better.
Is there something that DEC and Berkeley are supposed to do to their
sendmail.cf? Would it be possible for the implementors of smail to
write this down and tell the respective postmasters? Or do the
postmasters know it already and reject the smail way of doing things?
Who's right, a site like decwrl that has to actually move the mail, or
the smail implementors. Or is the name server in need of an update?
Or is this extra hop just part of the smail design? Will sun go nova
because of all these extra hops? What's going to happen to the speed
of my mail delivery as everyone starts pouring all their ARPA mail
through sun?
Here's the test mail:
From kbh@ucbvax.Berkeley.EDU Fri Jan 9 00:30:44 1987
Received: by amdcad.AMD.COM (4.12/smail2.1/11-19-86)
id AA18226; Fri, 9 Jan 87 00:30:39 pst
Received: from Sun.COM (arpa-dev) by sun.Sun.COM (3.2/SMI-3.2)
id AA25990; Fri, 9 Jan 87 00:05:09 PST
Received: from ucbvax.Berkeley.EDU by Sun.COM (3.2/SMI-3.2)
id AA03324; Fri, 9 Jan 87 00:05:18 PST
Received: by ucbvax.Berkeley.EDU (5.53/1.20)
id AA25107; Fri, 9 Jan 87 00:05:54 PST
Date: Fri, 9 Jan 87 00:05:54 PST
From: kbh@ucbvax.Berkeley.EDU (P. Ngai)
Message-Id: <8701090805.AA25107@ucbvax.Berkeley.EDU>
To: phil@amdcad.amd.com
Subject: testing
Cc: kiefer@decwrl.dec.com
Status: RO
Does ucbvax do it right? Or will this get routed through
sun like decwrl does?
Phil Ngai
--
Phil Ngai +1 408 749 5720
UUCP: {ucbvax,decwrl,hplabs,allegra}!amdcad!phil
ARPA: amdcad!phil@decwrl.dec.comdave@rsch.wisc.edu (Dave Cohrs) (01/10/87)
Phil Ngai complains that his mail gets routed through SUN.COM
despite the fact that his host is connected directly to other
sites, for example UCBVAX.BERKELEY.EDU and DECWRL.DEC.COM.
Without looking at the mail config files on the sites you mention,
I can't give an exact answer. However, the two sites you mention
*are* on the Internet and use the Nameserver for routing mail.
When they get mail for AMDCAD.AMD.COM, I assume that they look up the
mail routing info in the nameserver, an MX record, which says "send
*mail* for AMDCAD.AMD.COM to SUN.COM".
If amdcad (or any other AMD machine, as the case may be) has a direct
connection to decwrl and ucbvax, they need to put special hacks into
their configuration to override the MX information from the
nameserver.
In any case, I don't think your beef is with smail, but with the mail
configurations at decwrl and ucbvax. Decwrl and ucbvax are doing what
AMD.COM told them to do, "send our mail to SUN.COM". I'm sure that you
can work something out with their postmasters.
Dave Cohrs
(608) 262-1204
..!{harvard,ihnp4,seismo,topaz}!uwvax!dave
dave@rsch.wisc.eduphil@amdcad.UUCP (Phil Ngai) (01/11/87)
In article <3108@rsch.WISC.EDU> dave@rsch.wisc.edu (Dave Cohrs) writes: > > Phil Ngai complains that his mail gets routed through SUN.COM > despite the fact that his host is connected directly to other > sites, for example UCBVAX.BERKELEY.EDU and DECWRL.DEC.COM. > >In any case, I don't think your beef is with smail, but with the mail >configurations at decwrl and ucbvax. Decwrl and ucbvax are doing what >AMD.COM told them to do, "send our mail to SUN.COM". I'm sure that you >can work something out with their postmasters. AMD.COM never told them to send our mail to SUN.COM. It may be that most sendmails assume a .COM site is on the Internet and don't bother to check for a uucp connection and that the configuration files need to be changed. From what I hear, it appears that the details of this change are still under development. I would have to say that the release of smail was premature or incomplete. -- My left foot is digital, my right foot is analog. Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,hplabs,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com