kasper@arcsun.UUCP (Richard Kasperski) (01/13/87)
Help!!!
When I mail a message across the ethernet from a SUN to a Symbolics
the SMTP mailer on the SUN hangs while waiting for a reply to the
HELO message. It appears that the Symbolics never sends the appropriate
reply. However when a message is send from the Symbolics to the SUN
the transferal is succesful. This leads me to believe that the problem
is not with the Symbolics mailer but rather with the mailer on the
SUN( I have other evidence that points this way ).
On the net a while ago was a mail message about the 4.2 mailer not always
adhering to the SMTP standard. Something to do with not terminating lines
in the correct way? In an attempt to test this I have run the mailer with
the -ba ( arpanet mode ) which is supposed to terminate lines correctly.
We run the sendmail configuration file that was supplied to us by SUN
with modifications to the address translation scheme to handle our local
network of machines. The effect of these modifications can be ignored because
the machines only set up the communication path but never send any information
across it.
Any help would be greatly appreciated.
Richard Kasperski
Alberta Research Council
Calgary, Alberta
...{ alberta }!calgary!arcsun!kasper
PS. Asking for help when you feel you've missed the obvious is humbling!!!
merlin@hqda-ai.UUCP (David S. Hayes) (01/20/87)
In article <145@arcsun.UUCP>, kasper@arcsun.UUCP (Richard Kasperski) writes: > > When I mail a message across the ethernet from a SUN to a Symbolics > the SMTP mailer on the SUN hangs while waiting for a reply to the > HELO message. It appears that the Symbolics never sends the appropriate > reply. However when a message is send from the Symbolics to the SUN > the transferal is succesful. This leads me to believe that the problem > is not with the Symbolics mailer but rather with the mailer on the > SUN( I have other evidence that points this way ). I experienced the same problem about 8 months ago. I finally found that the Sun was not terminating its lines with a CR, while the Symbolics was waiting for one. I've modified my sendmail.cf to have two different ethernet mailers, as shown: ############################################################ ##### Ethernet Mailer specification Mether, P=[IPC], F=CDFmMnRsuX, S=11, R=21, A=IPC $h S11 R$*<@$+>$* $@$1<@$2>$3 already ok R$+ $@$1<@$w> tack on our hostname S21 # None needed. ############################################################ ##### Symbolics Mailer ##### Send to local lisp machines, with lisp domain rewriting. Msymbolics, P=[IPC], F=CDFmMnRsuX, S=12, R=22, E=\r, A=IPC $h # Sender rewrite. S12 R$*<@$-> $1<@$2.autocross> R$- $1<@$w.autocross> # Recipient rewrite. S22 R$*<@$-> $1<@$2.autocross> R$- $1<@$A.autocross> The determination of which mailer to use is controlled by ruleset zero. It has been given a list of Symbolics machines, as a class. A rule has been added: # Delivery to one of the symbolics machines. R$*<@$=A> $#symbolics $@$[$A$] $:$1 This specifies the symbolics mailer for any machine in class A, which is defined at the top of the cf file. The square-bracket construct in the right-hand side causes the lookup of the given hostname in /etc/hosts, and its replacement with the formal hostname. The domain rewrite rules were necessary at our site because the Symbolics identifies themselves as being in a different domain than the Sun system. Hope this solves your troubles, -- David S. Hayes, The Merlin of Avalon PhoneNet: (202) 694-6900 ARPA: merlin%hqda-ai.uucp@brl-smoke UUCP: ...!seismo!sundc!hqda-ai!merlin