E.B.:
> Hello,
>
> I
> wanted to allow certain clients to relay by using a check_client_access
> lookup map. It works nice if I use IP addresses. If I use domain names,
> it stops working for my test environment. My test client doesn't have
> rDNS set up (I think this is the cause of "connect from
> unknown[x.x.x.x]", right?).
>
> Do the check_client_access domain
> checks require rDNS?
Any decision based on SMTP client hostname requires either an entry
entry in /etc/hosts or equivalent, or a PTR record in DNS.
Of coure for access control Postfix also requires that the name
resolves to the client IP address, otherwise cheating would be
much too easy.
> Is there any way to make Postfix do a normal A record resolve on
> a domain I give it in that table? For my relay clients, I can trust
> their DNS A records and I can't think of why I need to check their
> rDNS (unlike untrusted straingers).
Postfix determines the SMTP client name first, and then it looks
up that client name in the access table.
Postrix does not iterate over the whole access table and find out
if any of those names resolve to the client IP address. If you want
that, then you can write a policy service plugin, and see how well
the idea works.
Wietse
