🚀 go-pugleaf

I've long been running MX 2.0 on a Vaxcluster, all seven of whose nodes are
currently running VMS 5.5 and CMU-TEK version 6.6-5.  I haven't upgraded MX
for two reasons:  1) "If it ain't broke, don't fix it" and  2) A couple of
previous attempts to upgrade to 3.1 have been unsuccessful.  However, a recent
lack of success on my users' part in sending messages to the CERN particle
accelerator site in Europe (addresses ending in .CH, e.g., joe@foo.bar.ch; the
messages simply vanish and no bounced mail messages are received) is forcing
me to confront this upgrade (to 3.1; unless somebody advises me otherwise,
I'll get 3.1 working first and then worry about 3.1C or whatever is the
current revision) and conquer it by any means necessary, in hopes that the
newer version will have less trouble sending to CERN.

Here's my problem:  I've customized the 3.1 installation for our site (the
ROUTER, LOCAL, and MLF processes all run on 3 of the 7 nodes; SMTP runs on
these 3 plus one other node; and SMTP_SERVER runs on all 7 nodes).  As best
I can tell, I've modified it so it should be configured just as it has been
to successfully run MX 2.0.  I start up MX and all the appropriate processes
start up, and will run until I stop them myself.  However, as soon as I try
to send an outgoing message, I receive the following error:

MAIL> send
To:     mx%"user@site.edu"                 ! dummy user and site, of course
%RMS-F-DEV, error in device name or inappropriate device type for operation

I've checked all of the appropriate logicals I'm aware of (SHOW LOGICAL *MX*
and SHOW LOGICAL *NETLIB*), and as best I can tell they're all defined
properly; i.e., all logicals defined to point to directories or files do so
successfully and MX_INET_HOST, MX_NODE_NAME, MX_SMTP_DEFAULT_ROUTER and
MX_VMSMAIL_LOCALHOST all point to the correct machines.  Does anybody have
any idea what could be wrong here?  Apologies if this is a beginner's question;
I've recently taken over MX duties from the guru who set up 2.0 on this site
and am still on the learning curve.  Also, please let me know if I can supply
additional useful information.  Many thanks in advance.

- Patrick Walsh
  University of Virginia Department of Physics
  pw@virginia.bitnet, pw@virginia.edu, pat@gomez.phys.virginia.edu