>So, if a host 1.1.2.3 connects, and presents a envelope sender of
>[EMAIL PROTECTED], postfix does an rdns lookup on 1.1.2.3

... gets the PTR hostname,  the queries for it A record to see it matches 
1.1.2.3

>If that rdns fails, the unknown_client_reject_code is sent and the mail
>is rejected.

right

>OTOH, if 1.1.2.3 returns a PTR of (for argument's sake)
>mail.badbadbad.org, then postfix will do a forward dns lookup on the
>returned name (mail.badbadbad.org).

right and if the A record of mail.badbadbad.org is 1.1.2.3, it's a match.

>If mail.badbadbad.org has NO A record, postfix rejects the mail.

right

>If it does have an A record (or multiple A records), one of the A
>records returned must match the MTA IP.

right

>So, in my case, my server sending out mail from 198.235.200.78 has these
>two records:
>
>         78.200.235.198.in-addr.arpa     name =3D mail-byron.theedge.ca
>and     mail-byron.theedge.ca   internet address =3D 198.235.200.78

so A and PTR "match"

>Theedge.ca is my provider, and I *never* send mail from anything at
>theedge.ca. (But, I'm not a commonly forged domain :) ). So, therefore,
>my mail passes the reject_unknown_client test.

right

>What I don't understand, is how you enforce the fact that the PTR must
>be in the same subdomain as the returned lookup.=20

A and PTR must match, there is no other requirement. The PTR hostname can 
be anything.

Len


Reply via email to