Wietse:
> No matter what MTA you use, it will need to know a) how many the
> sender has sent and b) what the limit for that sender is.
>
> Therefore, some per-sender configuration is unavoidable.
Steve Fatula:
>True of course, but, was thinking of using a milter to do this since
>it works for locally submitted mail as well. However, your solution
>makes way more sense. The milter or other methods block mail if you
>exceed some limit. Using your way, they simply pile up in the queue
>I presume, which means software still works and is not blocking
>anything. So, mailing list managers like mailman still work (they
>don't have rate throttling unless the newest version does). I like!
It does indeed pile up. Unfortunately the excess will pile up in
the active queue, which is shared. Once that queue is full, things
slow down for everyone.
A more serious issue is that _destination_rate_delay is per-destination
not per-sender, so the example that I gave was wrong to begin with.
This would require a rate delay that is independent of destination.
Wietse
