We run a number of backup MXs for our customers and I'm doing some
tuning to reduce the number of mails that languish in the queues on
these machines. A large number of the mails stuck in the queues are
spam to non-existent addresses at the third party server. At the moment
the mails sit in the queue being retried periodically and are then
finally deleted after a week. I would like to reject these mails much
sooner.
Looking through the ACL documentation I see the callout verification
mechanism that looks like it will do the job, but I have a number of
concerns:
- When a callforward is performed is the sender's connection to my
server kept open? My guess is not.
- If a check is successful is the message sent over the same connection?
- What does the sender see when the callforward check fails
(non-existent recipient)? My guess is that this depends on the way I
write my ACL?
- What happens when multiple recipients are specified? Is a callforward
tried for each recipient?
