Sendmail question...

[Dan Ts'o]

	I'm just starting to learn my way around sendmail (does every 4bsd
site have to go through this ?) and stumbled across a problem that I can't
figure out, short of plunging into full-time debugging.

	We have a local ethernet, and our main machine, RNA, has UUCP
connections and ARPAnet access via the Rockefeller campus machine. I got just
about everything to work (including using smail/pathalias for UUCP routing)
except multihop forwarded local mail.

	If I make my personal mail machine RNA, and my accounts on say, RNB
and RNC have .forward files that redirect mail to dan at rna, if I mail to
myself from RNB to RNC, RNA's sendmail says:

	rna: I refuse to talk to myself (unavailable service).

	Apparently in the process, the destination address is rewritten to
dan at rna.rockefeller.edu (which seems fine). "Single" hop local forward works
fine (e.g. from RNB to RNA). Just mailing from any machine to
dan at rna.rockefeller.edu works fine. Everything works fine except this multihop
forwarding.

	I know the message is in the sendmail code that tries to open a TCP
connection, but I can't understand how RNA's sendmail ever gets there. In the
sendmail.cf, there are rules to catch local mail and use the local mailer. In
desperation, I commented out in sendmail.cf every instance of $#ether and
$#tcp, as well as those mailers, the only mailers that use the IPC channel, and
still the problem occurs exactly as before.

	I therefore concluded that there was some other way that sendmail was
deciding to use IPC/TCP instead of local mail, bypassing the sendmail.cf
completely. Am I all wet ? What am I missing ?

	Thanks.


				Cheers,
				Dan Ts'o
				Dept. Neurobiology	212-570-7671
				Rockefeller Univ.	...cmcl2!rna!dan
				1230 York Ave.		rna!dan at nyu.edu
				NY, NY 10021		tso at rockefeller.edu