Using sa-learn to add ham mail to the spamassassin database

Antony Stone Antony at SOFT-SOLUTIONS.CO.UK
Fri Jul 4 10:33:02 IST 2003

On Friday 04 July 2003 10:16 am, Plant, Dean wrote:

> I am using MailScanner 4.21-9 on Redhat 8 with spamassassin 2.55, dcc and
> razor2. MailScanner is used as our external relay with all internal mail
> handled by exchange/outlook.
> For low scoring spam I am using the "attachment deliver" spam actions.
> My question is regarding false positives of low scoring spam and the best
> way to teach spamassassin with sa-learn.
> I had asked users to drag and drop incorrectly identified spam mail into a
> public folder but after the spam is delivered the spam mail is changed to
> have the subject line {spam?} with the original header information and the
> original mail is the attachment but the header info has been removed. So
> using this mail with sa-learn will not give the correct results.
> Can anyone advise the best way to get the original mail with all the header
> info back onto the MailScanner server.

You could change the actions for low-scoring spam to include 'quarantine',
and then use a script to pick the message ID out of the false-positives which
users drag and drop into the public folder, match that with the filenames of
the quarantined messages, and there you have the original false-positive
message with no changes?




What is this talk of software 'release' ?
Our software evolves and matures until it becomes capable of escape,
leaving a bloody trail of designers and quality assurance people in its wake.

More information about the MailScanner mailing list