/etc/mail/relay-domains vs. access database RELAY

[Brandon Hutchinson]

Hello,

I have a question regarding the difference between using a relay domain
as an entry in /etc/mail/relay-domains ($=R) or as a key in the access
database.

First of all, I am using FEATURE(relay_hosts_only). I am not using
FEATURE(accept_unresolvable_domains). The tests were performed on a
sendmail 8.13.5 installation.

If "example.com" is in /etc/mail/relay-domains, sendmail will
rightfully reject any non-existent subdomains of "example.com" in the
check_mail ruleset. Example: MAIL FROM:
<us...@nonexistent.subdomain.example.com> would return "553 Domain of
sender address ... does not exist"

If "example.com" appears as a key in the access database with a RHS of
RELAY, any sender address ending in ".example.com" is acceptable.

Is this expected behavior? According to
http://www.sendmail.org/tips/relaying.html:
Keys can be a fully or partly qualified host or domain name such as
host.subdomain.domain.com, subdomain.domain.com, or domain.com. The
last two forms match any host or subdomain under the specified domain.
(If FEATURE(relay_hosts_only) is set, only the first form works.)

In this case, the key is domain.com (example.com). Since I'm using
FEATURE(relay_hosts_only), does that mean that any host or subdomain
under the specified domain should not match?

Any help would be greatly appreciated!