Q:  I'm sending mail via UUCP.  Things work great from my mail server (which is also my UUCP gateway).  From a mail client, though, I get error messages indicating the remote UUCP machine is unknown.  The bounce message looks like this:

Date: Mon, 13 Apr 92 13:11:11 CDT
From: Mailer-Agent (NeXT Mail Agent)
Subject: Returned mail: Unable to deliver mail
To: amm

----- Transcript of session follows -----
554 rhino!sandy... Never heard of UUCP host rhino

----- Unsent message follows -----
...

What's going on here?

A:  One likely source of the problem is that sendmail might not have an uptodate notion of the known UUCP connections.  If you've added the remote machine (rhino in the example, above) and neither rebooted the server nor restarted the sendmail process on the server, then sendmail doesn't know about the new connection.

A sendmail daemon process is started when the system boots, by /etc/rc.  When sendmail starts up, it runs the command uuname to determine the known UUCP connections.  (At least, it does this with all the standard configuration files NeXT ships.)  The uuname command is run only once.  It is this sendmail daemon that handles all the mail coming from the client machines.

So, why does mail originating locally work?  A new sendmail process is run for each locally-originating message; this new sendmail process runs its own instance of uuname when it starts.  Since uuname for the new sendmail process has been run after the new system's addition to L.sys, the new sendmail knows about the new system.

QA835

Valid for 2.0, 3.0, 3.1