The documentation here suggest there are a three of ways an email is eligible for signing.
an email must either be:
A) received from an authenticated user,
B) a reserved (local) IP address, or
C) an address in the sign_networks map (if defined).
I have mailcow
configured for the domain example.com
. It is an internal DNS server, so the mx
record correctly points to the SMTP server the email gets routed to when it arrives from an external source. Let’s call it internal_mail.example.com
.
internal_mail.example.com
is also used to send email. It connects to the mailcow
server as unauthenticated hands the email over to mailcow
to send.
Mailcow
does not have a user list, so it doesn’t know that the user is authenticated. User mailboxes are contained in the internal email server.
This appears to dismiss option A
above.
So I tried option C
. My dkim_signing.conf
has a bunch of defaults set. One of those is
sign_networks = "/etc/rspamd/custom/dovecot_trusted.map";
. I tried updating dovecot_trusted.map
with the valid internal networks that would be connecting to mailcow
to send email. However, I noticed that dovecot_trusted.map
removes my manual additions and reverts to a couple of addresses it puts there automatically.
So that appears to no longer be an option.
How do I use option B
?
When mailcow is being used as purely as an MTA relay with a spam filter, how does one get it to sign email that it relays for our internal mail server?
Thanks
Kevin