Rosenwald, Jordan wrote:
I agree structure = good, but in this case what is the endgame? What's
the value? What would you do with the information if MTAs returned it?
Teach the MTA when they should reschedule to not waste attempts.
When we first implemented GL in early 2003, the two immediate needs were:
- Shorter 2nd and 3rd Retries (within 5 mins)
- Automated whitelisting of customers
So our implementation comes with auto white listing of recipients you
directly write mail to with the idea you know them and would expect
responses. This is done at the SMTP receiver for authorized mail
authors of the local host. When anonymous senders, the greylist
process checks this auto-whitelist.
And shorting the initial retries help address the delivery delays
against remote sites now implementing Greylist.
While the history has been there were systems that wanted you to retry
much later, maybe 1 hour, what I have been seeing increasing it has
beyond 1 hour into days now. Today, this has caused some false
positive bounces and flagging of "bad destination" address.
What will help is a structured response that will help the MTA to
reschedule at the appropriate time and now waste all the attempts.
Further, there are customers with large mailing list to different
destination email domains but are hosted by the same remote host.
i.e. many email domains dealing in the same trade might be using the
same set of MX hosting domains. They are now greylisting too and
there are now a tremendous amount of waste in sending attempts.
Some don't even accept mail until after 5:00pm - end of the business.
Some TAKE down their server at 5:00pm and only bring it back at 9am
the next morning.
I am seeing far too much waste in sending mail with GREYLIST
implementations or those that "behave" like so with non-FULL time
operations.
Something can be done.
--
HLS