From: Noel Jones (njonesmegan.vbhcs.org)
Date: Wed Dec 10 2008 - 09:59:49 CST

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Dennis // supportpowerhosting.dk wrote:
> Hi all.
>
> I manage a few postfix spamfilter gateways for my company.
>
> They are doing a really god job of scanning mails and delivering them to
> customers mailservers.
>
> The way we implement this is using a catch-all address to accept all
> mail for customers domains, then scan the mail and try to deliver it to
> the customers mailserver.
>
> This is of course a less than ideal solution, as i´m currently accepting
> a lot of mails for remote-addresses that are non-existing. Which means a
> few 100k mails scanned every day just to have the remote server bounce
> it as the recipient does not exist.

Right. If you send enough backscatter, you'll get blacklisted.

>
> I then thought of using the "reject_unverified_recipient" parameter,
> thus letting the spamscanners probe the remote mailserver to see if the
> recipient is valid or the mail would bounce.

A fine solution if you are unable to get an actual list from
the remote mailservers.

>
> My questions are:
> As the spamscanners are the best or primary MX´s
> for the customers domains, would postfix then just probe itself, and
> always get a positive answer due to my catch-all entry ?
>
> Or would postfix actually look at the transport map, and realize that it
> has to probe the customers server. ?

Postfix uses its normal routing methods with verification
probes. If "normal" mail goes to the right place,
verification probes will use the same channel.

>
> And how would one go about controlling the destination of the
> probe-messages ?

You probably don't need this, but here are the docs for probe
routing:
http://www.postfix.org/ADDRESS_VERIFICATION_README.html#probe_routing

--
Noel Jones

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]