I am looking at the new SMTP callback stuff, and I realize it is expensive
in terms of network and whatever. We run virtual mail hosting for a number
of domains (expected to grow when we go commercial) and I would like to be
able to optionally provide spam-trap / checking services.
The experiences and advice of the list (wrt current best practice) would be
appreciated.
What I have thought of doing, through a customer configuration interface, is
to change the lowest MX record for those domains that want it to go via a
separate filter system; so for example where knowledge.com now has a lowest
MX of mailstore-1.XXX I would change this to spamtrap-1.XXX.
On this machine, what options should I go for (I am taking the all or
nothing approach I think for simplicity). I would like to use SMTP
callbacks, some RBL stuff and so on. What do people currently use as a
realistic, pratical set of configs for maximum cleaning power ?
Then, this machine would privately relay any successful mail back through to
the real mail storage machine...
Can anyone see a way of providing this type of thing, but for specific user
addresses per domain or is this easier done on a domain basis ?
I can see using some aliases (looked up via some external source) to
selectively choose to route some peoples (or mailing lists) mail via this
spam-trap machine ... but what do I need to watch out for when forwarding
the selected mail back into the delivery system (to avoid loops obviously) ?
rgds,
--
Peter Galbavy
Knowledge Matters Ltd.
http://www.knowledge.com/
