billd@fps.com (Bill Davids_on) (11/22/89)
This is a weird one and I can't seem to gleen the appropriate info from the RFC's. I am used to running sendmail on BSD systems and now I have to deal with a a VMS gateway running Wollogong (sp?) SMTP and it's giving me hell. Normally, when I send to xxx@yyy, sendmail connects to yyy and does a RCPT-TO:xxx. We send to the VMS gateway, vax11 this way: xxx%yyy.decnet@vax11 Sendmail does a RCPT TO:xxx%yyy.decnet. This is not acceptable to the Wollogong software. It wants to see RCPT TO:xxx%yyy.decnet@vax11. Wollogong claims that the local machine is required and that I'm wrong to send things the way I do. Who's right? Can someone please point me to the appropriate pages of the appropriate RFC's (I've been going through 821 and 822)? I would think that vax11 should know who it is (I know who I am :-) and not need the extra addressing. It seems unnecessarily redundant to me. --Bill Davidson
billd@fps.com (Bill Davids_on) (11/23/89)
In article <4173@celit.fps.com> billd@fps.com (Bill Davids_on) writes:
[desciption of problems talking to Wollongong SMTP]
Thanks to all who have replied so far. As soon as I resolve this
I will sumarize my solution. A couple of people noted that I sent
"RCPT TO:xxx%yyy.decnet" to the Wollongong SMTP and said that
it needed an "@" symbol somewhere. Well, I tried that. In fact,
I connected to it with telnet and fed it about 10 different permutations
of addresses and the only thing it would ever take was
"RCPT TO:xxx%yyy.decnet@vax11". It hated everything else. It
*required* the "@vax11" on all addresses.
I've also gotten a several comments from people who hate Wollongong
software. Here's a sample:
One person refered to it as "Will-go-wrong SMTP"
Another said: "Always assume that Wollongong is wrong. They read
RFC's like the Devil reads the Bible."
From the same person: `"Be liberal in what you accept and conservative in
what you send." (Jon Postel, as quoted in RFC1122/3) This is called the
Robustness Principle. Wollongong can't even spell "Robustness"'
It looks like I have my work cut out for me. One person who had
the same problem suggested a special rule in sendmail.cf. I
can do that but I have to do it to every machine on our network
(about 200 machines). I don't relish the thought much but this
may be the way I go (actually, I'll probably only do those
machines where people who talk to others on VMS normally work).
Another person suggested using MX records. I'd have to upgrade
sendmail to do that. I should do it anyway but I have a lot
of other priorities right now and going to a new sendmail has
been non-trivial in my experience.
Some other RFC's have been suggested as well (876,1122,1123). RFC
876 paragraph 5.7.2 apparently supports Wollongong.
Thanks to the following people for responding and giving suggestions:
andy@gollum.hcf.jhu.edu (Andy Poling)
lindberg@cs.chalmers.se (Gunnar Lindberg)
jack@piring.cwi.nl (Jack Jansen)
enag@ucsd.uucp@ifi.uio.no (Erik T Naggum)
isavax!cliffb@uunet.UU.NET (cliff bedore*)
parmelee@wayback.cs.cornell.edu (Larry Parmelee)
If you replied and you're not listed then I didn't get your message
(as of 5:25pm PST Nov 22).
--Bill (I *wish* I was a sendmail guru) Davidson