On Wed, Aug 16, 2023 at 01:51:24AM +0200, Étienne Miret via Postfix-users wrote:
> I found this discrepancy surprising and am suggesting it is removed. In
> case others argue it is useful or that removing it will break some
> configurations, I am asking it is documented.
The discrepancy is intentional. Aside from virtual(5) which is
definitive for the otherwise "fictional" virtual alias domains
(have no real recipients), for the remaining address classes the
queue manager just hands off the message to the relevant transport.
Whether the transport also uses the same table to decide whether or
where to deliver mail is up to the transport.
For the virtual mailbox address class, the built-in virtual(8) delivery
agent does sue virtual_mailbox_maps, but anoother VDA might not.
The relay transport will just deliver mail without regard to
relay_recipient_maps, and local(8) will use a variety of strategies to
try to deliver messages.
The various recipient tables are a moat to fend off hostile invaders,
but play no explicit rôle in delivery of messages that make it inside.
--
Viktor.
_______________________________________________
Postfix-users mailing list -- postfix-users@postfix.org
To unsubscribe send an email to postfix-users-le...@postfix.org
