It wasn't possible to offer an opt out as the changes were to the Exim configuration on the servers themselves and were applied across the board without an API to handle any sort of exclusion.
Bob, I have an idea about the mail server platform that may be food for thought.
I wonder if you could configure the load balancers to direct mail to specific relays based upon the ACL procedure you put into Exim and have subsequently rolled back.
Consider two groups of relays:
1. "Aggressive" relays, which provide a full DSPAM and CLAM filter
2. "Passive" relays, which have a reduced DSPAM configuration
When an inbound mail hits the load balancer, it performs a (fast) check on your dynamic IP ACL. If it matches, then it passes the mail off onto the aggressive relays. If it doesn't match, it is then sent through the more passive relays.
The thought behind it is that the majority of spam originates from dynamic IP addresses. As long as you have the capacity to process this 'spam' email on the aggressive relays, that leaves the passive relays more time to process their (likely genuine) mail more expediently.
This would, however, put more load on the load balancers, if they can handle this sort of ACL check at all!
It's a bit like QoS for Email - those mails *fairly likely* to be spam could potentially be delayed a short while, whereas mail that is *fairly unlikely* to be spam will get processed by the faster 'passive' relays.
That way, all messages are processed, none are bounced with a 550, and you can start to identify methods on the aggressive relays with which to further fine-tune the process.
You could even tag the mail passing through each relay "type" with an: X-Header identifying 'Passive' and 'Aggressive'. Personally, I would be more than happy to join a trial whereby any mail that was misrouted could be sent to
abuse@plus.net for further investigation.
This would help you to generate your 'whitelist', with the only customer impact being the *possible* delay of mail for a short while.
Thoughts?
B.