Am 29.09.2015 um 23:45 schrieb coolhandluke:
based on just what i've found in the last 10 minutes, i would be very
careful about scoring anything related to {invalid|missing|extra}
headers too high. definitely test your rules extensively (with very low
scores) before rolling them out to production!
you need to train your bayes at the same time and raise the neagtive score of BAYES_00-BAYES_20 - a spamfilter is always a adaptive system
here the combination if missing MID/DATE with a good bayes is never a problem and in combiantion with a high spam-byes core a clear sign for mails to reject (yes all spamass-milter rejects are reviewed careful)
signature.asc
Description: OpenPGP digital signature
