Spam Scoring issue

Tom Hendrikx tom at whyscream.net
Tue Aug 26 20:09:29 CEST 2014


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 25-08-14 21:25, Vernon Fort wrote:
> I am testing messages that made it pass the amavis scanner with a
> score less than 1 (0.1333).  But when I sync the message from an
> exchange folder and test is manually by ‘spamassassin –D < {MID}.’,
> I get a score above 16.  These messages should have scored
> initially very high but they did not.  Where do I start looking?
> 
> 
> 
> Vernon
> 

Messages in Exchange have virtually all headres munged or deleted.
They are useless for debugging since they don't resemble the original
message at all. I bet the 16-ish score includes spamassassin rules
named MISSING_<foo> and such? It could also be related to blacklists
not hitting during initial delivery. Maybe you could share a sample
with us (and/or a spamassassin debug output)?

If you need debugging on those messages, start by setting up a tap for
all messages, so you don't have to pull the samples from exchange. In
postfix, you could use 'always_bcc' (or similar) to save a copy
somewhere outside of exchange. See
http://www.postfix.org/ADDRESS_REWRITING_README.html#auto_bcc for
examples.

Tom
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=aEJh
-----END PGP SIGNATURE-----


More information about the amavis-users mailing list