Marc Chametzky wrote:
I can't get over this nagging feeling in the back of my head
that it opens up a customer to a new variety of DoS attacks
What we're discussing at the moment is to limit the worst case
to about 10..25 queries, with classic-spf-00 it's still 110
(10 mx or ptr, each with 10 names).
Radu also proposed a way to identify malicious PermErrors by
going over the limit, and use a higher blacklist limit. But
that's not necessarily a good idea, an attacker could stay
below the higher BL limit.
Bye, Frank