spamassassin sa-learn

[Glenn Steen]

On 9 December 2010 18:39, Peter Ong <peter.ong at hypermediasystems.com> wrote:
> Hello Everyone,
>
> I'm trying to do an sa-learn.
>
> There is an email that went through MailScanner. It was spam, but spamassassin did not classify it as that. It's from webmaster at greeceiloveyou.com. It arrived into the inbox of one of my users.
>
> I took that email in its original form after it had been delivered into his inbox. Therefore, this email contains his email info on the top. If I sa-learn the spam that contains headers with his email, does spamassassin begin to think emails with that person's address in the header is spam?
>
> What would be the best way to sa-learn new spam? Should I delete headers containing the emails of people who received it first? Thanks.
>
> p

Although this is pure SA, searching the list archives for things like
ALL_TRUSTED should lead to some fairly detailed information on the SA
trust path and what it does for you... You could look it up in the SA
wiki as well... Also look at you /etc/mail/spamassassin/mailscanner.cf
file, and check that you've made all needed configs there.
If all that is OK, sa-learn shouldn't have a problem with a locally
delivered message. Of course, it'd be easier to work with the MS
quarantine, so some actually keep all non-spam messages in there for a
day or two, just to be able to do easy learning (from MailWatch, for
example)... Just a little scripting to clear it out...:-)

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se