MailScanner/SpamAssassin VS spamassasin -t gives different scores (sometimes)

[Peter Bates]

Hello all...

> dot at DOTAT.AT 24/04/03 15:09:11 >>>
>Which rules? If you run it in debug mode, what does it say about auto-
>whitelists and bayesian analysis?

Also see my message from yesterday (23rd) entitled 'SpamAssassin Weirdness' that got entirely lost in other discussions ... basically I'm seeing something similar with SA 2.53, and the latest beta of MS, using Postfix support (but I don't think the MTA has much to do with it)...

I'd appreciate a bit more insight... also 'Debug SpamAssassin = yes' dumping some stuff when MS is actually running as a daemon would be quite nice. Playing devil's advocate, amavisd has some really nice timing statistics as to time spent processing each message, and how much time (and percentage of time) is spent on certain elements, e.g. SA took 79% of the whatever milliseconds the complete mail took.

On another note, Bayes starting working for me with the above combination yesterday, and I haven't a clue why, as I hadn't changed anything... suddenly bayes_msgcount started incrementing, when it hadn't been before... crazy!



--------------------------------------------------------------------------------------------------->
Peter Bates, Systems Support Officer, Network Support Team.
London School of Hygiene & Tropical Medicine.
Telephone:0207-958 8353 / Fax: 0207- 636 9838