> And what I'm thinking is that if we offer it as a strictly store and
> forward service, we end up with way too much spam on our servers and we
> do clients a disservice because they can't bounce spam on "rctp to".
I have a secondary setup, but it's really more designed as a fallback --
i.e. if the primary dies I can still send and receive mail. It's
actually on a machine that is almost a mirror of the the primary so it
can replace the primary if needed.
accept domains = +relay_to_domains
endpass
message = unrouteable address
verify = recipient/callout=20s,defer_ok
As I said, the secondary is really a standby mirror of the primary. It
has almost exactly the same exim config, users, aliases and so on as the
primary. So the secondary really doesn't need to make the callout to
the primary because it already knows enough to decide if the mail
should be accepted.
But since my relay_to_domains router is at the start of my routers it
can't verify locally. I guess I'd need to have some way of overriding
the transports so everything is relayed to the primary after routing
checks. But I have not spent any time thinking about it yet -- and I'm
not even sure it's possible without replacing all the transports (and my
limited knowledge of exim).
--
Bill Moseley
moseley@???
