One of the systems I manage runs RANCID and when we recently changed the method of procedure as far as who the system notifies for device diffs, I stumbled onto a problem I've yet to fix the right way.
RANCID uses two users in /etc/aliases per group (as created as part of a conf file):
rancid-admin-GROUP: rancid-GROUP rancid-GROUP: noc noc: user@fqdn.com The procedure we changed required we send to users that are on the same domain as the server. This change broke outgoing messages as the server now complained that the user is unknown to the system:
Aug 25 10:01:02 <deleted> postfix/local[5684]: 0826CABB7: to=<user@fqdn.com>, orig_to=<rancid-admin-GROUP>, relay=local, delay=0.2, delays=0.12/0.02/0/0.06, dsn=5.1.1, status=bounced (unknown user: "user") I've attempted using various methods such as virtusertable, relay changes and... (and forgive me I'm drawing a blank on the others as I tried so many individual changes), but never had any successful changes that both corrected the "unknown local user" problem while maintaining the use of /etc/aliases as needed by RANCID.
My current bandaid fix was to change the server's domain to one of our other domains. This works, but e-mail sent now shows the local group name (from /etc/aliases) in the recipient lists instead of the true external-to-this-system user (and the from address shows the incorrect domain).
Added note: this box does not need to receive any mail from external sources and is intended to be a outgoing only 'no-reply' type system.
Edit: This is fixed good enough to my liking; Details in comments.
[link][17 comments]