Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email email@example.com
From: Peter Thomassen (mailpeter-thomassen.de)
Date: Sat Jun 09 2007 - 04:09:14 CDT
Wietse Venema wrote:
> Peter Thomassen:
>> Why? If the score is at least 3.0, the message is processed to the
>> greylisting routine that decides about temporarily rejecting (tuplet
>> unknown) or accepting the message (tuplet known). I don't see how that is
>> broken by adding SpamAssassin to pass by messages that are unlikely to be
>> spam (score < 3.0).
> 1) Good MTA sends low-score mail. No greylisting, no greylist tuple.
> 2) Same good MTA sends high-score mail. Do greylisting, no greylist
> tuple, useless delay.
It is unlikely that a good MTA sends high-score mail: When watching the
greylisting log on my Exim installation (where such conditional greylisting
is implemented) I can see from the headers stored there that almost no
high-score ham comes in.
> With MTA auto-whitelisting it impacts regular MTAs hardly.
Of course I already use auto-whitelisting. As I said, in my greylisting log
there is almost no ham. If greylisting didn't depend on the score, every
unknown tuplet got greylisted, and I think that the tuplet of about 15% of
all ham is unknown (in the last month). I think that there are less
high-score hams (whose tuplets than are greylisted on first occurence) than
unknown ham tuplets altogether.