grv@tahoma.UUCP (Jerry VandenHeuvel) (08/20/88)
We have three Apollo rings with a gateway node on each ring connected to ethernet, each gateway are running sendmail. We also have Alliant, Sequent and Motorola Unix systems on the same ethernet. Mail sent from any Apollo node to the other systems has the from address of user@node, not user@gateway-node, so replies have to be manually changed before mail transfer will work. This same problem applies to any incoming mail to an Apollo ring, forcing users to know which node is configured as the gateway. Is there a way to make the sendmail configuration files on the other Unix systems smarter by rewriting an address for any node to the gateway nodes? I have the basic sendmail documentation and the example configuration files which come with Berkley distribution, none of which are too helpful. I would appreciate any hints, suggestions or examples (!) that anyone might be able to send via email. -- Jerry Vanden Heuvel Voice: (206) 237-5080 Boeing Commercial Airplanes UUCP: ..!uw-beaver!ssc-vax!shuksan!tahoma!grv P.O. Box 3707, M/S 66-04 Seattle, WA 98124-2207
geof@imagen.UUCP (Geoffrey Cooper) (08/23/88)
In article <261@tahoma.UUCP>, grv@tahoma.UUCP (Jerry VandenHeuvel) writes: > and Motorola Unix systems on the same ethernet. Mail sent from any Apollo node > to the other systems has the from address of user@node, not user@gateway-node, > so replies have to be manually changed before mail transfer will work. This > same problem applies to any incoming mail to an Apollo ring, forcing users to > know which node is configured as the gateway. Is there a way to make the > sendmail configuration files on the other Unix systems smarter by rewriting > an address for any node to the gateway nodes? I have the basic sendmail Here is what we did. In /usr/lib/sendmail.cf, add a line that explicitly defines "w" to be the name of the gateway. Normally "w" is the host name, so this has the effect of forcing the reply to be the gateway. We originally did this patch to avoid having replied mail go to diskless nodes that might not be up. We found we really like it because it centralizes the sendmail server for all the apollos on a ring (no one else has to run "sendmail -bd") which makes it easier to maintain. But that is because we have found a service that is at 100% service 90% of the time preferable to one which is at <100% service 100% of the time. # This line makes mail sent from any apollo come from # user@apolling. Cw apolling - Geof -- UUCP: {decwrl,sun}!imagen!geof ARPA: imagen!geof@decwrl.dec.com