[plug] Q for the spamassassin experts

[William Kenworthy]

On Tue, 2006-01-03 at 09:08 +0800, Bennett, Phillip wrote:
> Bill,
> 
> What you need to do is get all these spam messaegs into one place and in
> mbox format.  Then use the sa-learn command on them.  From memory, it's
> just 'sa-learn --spam --mbox <filename>' and it'll learn that these
> messages are spam and should block them in th future.
...

been doing this for many months (and its far more than just a basic
spamassassin setup, see
http://www.gentoo.org/doc/en/mailfilter-guide.xmletc)

Its maintenance info I need - do I keep adding suspect emails, and those
that I want rated as spam to the spam directory (which has sa-learn
running on it via cron), or do I remove older messages (say >3 months)
to avoid problems.  It seems to me that adding thousands of messages
over a long time period might create a skewed, or ineffective database.

What has brought this up is that the leakage of spam through
spamassassin is increasing as the spam writers are getting cleverer and
the spam is looking more and more like valid email.

BillK

-- 
William Kenworthy <billk at iinet.net.au>
Home!