Looks like bayes is dropping the score. Did you manualy train your bayes database, or use autolearning?
You may need to delete your bayes database and start over. On 5/4/05, BAKONYI Péter - paha <[EMAIL PROTECTED]> wrote: > Hi! > > >>I'm new at this list. I'm using spamassassin on my debian box since 2 > >>years and now I've upgraded to a new machine, new installation and a > >>new version (3.0.2) os spamassassin. But something seems to be > >>wrong because I got a lot of spam messages which is not identified to > >>be a spam because of some _negative_ points. I'm sure that they were > >>correctly identified to be a spam on my previous version on SA. Can > >>someone tell me any ideas? > > Can you post an X-Spam-Status header? > > Without that, no real clue, but I can guess that you might have had > > ALL_TRUSTED hit. If that's the case, read this: > > http://wiki.apache.org/spamassassin/TrustPath > > thanks for your replies. here's an X-Spam-Status header from a message > which was identified to be a spam before the new version and not spam > right now: > > X-Spam-Status: No, score=1.5 required=6.6 > tests=BAYES_00,DATE_IN_FUTURE_96_XX, > EARN_PER_WEEK,HTML_50_60,HTML_IMAGE_RATIO_06,HTML_MESSAGE, > HTML_WEB_BUGS autolearn=no version=3.0.2 > > thanks, > > peter > -- Ryan Castellucci http://ryanc.org/
