[Mimedefang] bayes filtering for a mimedefang implementation

[Kevin A. McGrail]

Matt,

Bayesian definitely works in a site-wide mode, autolearn process like you
are talking about.  However, it won't keep in on the tests until you have
like 200 or 300 messages in the learned queue.

Assuming you are running as a defang user, in the defang directory, you
should have a .spamassassin directory with your Bayesian files in there.  Do
you have the bayes_msgcount file or anything showing that it has started
storing tokens?  Has it gotten to 200 or 300 yet?

I have my spooldir in /var/spool/MIMEDefang so I also put my bayes files
there.  I used the following in my spamassassin config file to achieve this:

 #Enable bayes
   auto_learn 1
   use_bayes 1
   bayes_path  /var/spool/MIMEDefang/bayes
   bayes_file_mode 0666

KAM
----- Original Message ----- 
From: "Matt Cramer" <mscramer at armstrong.com>
To: <mimedefang at lists.roaringpenguin.com>
Sent: Friday, August 08, 2003 3:24 PM
Subject: [Mimedefang] bayes filtering for a mimedefang implementation


> I am running SpamAssassin 2.54 and MIMEDefang 2.35.
>
> Is it possible to use Bayesian filtering and bayes_auto_learn (in SpamAss)
> with MIMEDefang?  I have no local users on these mail servers, as they
> only relay mail to our internal mail system.  I have run sa-learn
> manually on some mail on our test system and the bayes files were created
> appropriately.  However when I send mail through the machine I do not see
> any changes to those files that would indicate auto learning is actually
> working.  I have put in some custom SpamAss rules and modified the default
> bayes auto learn threshold to ensure that my test message meets the
> criteria in the man page.
>
> I reviewed the last few months of list archives and saw one message saying
> that auto learning doesn't work with MIMEDefang:
>
> http://lists.roaringpenguin.com/pipermail/mimedefang/2003-May/005803.html
>
> but other messages indicating others have it working
>
>
http://lists.roaringpenguin.com/pipermail/mimedefang/2003-August/007235.html
>
> This of course begs the question of whether or not learning based on
> SpamAss's rating is a good idea.  However a manual sort is not practical
> given my users' skills, patience, and MUA.  ;)
>
> Any help would be appreciated.
>
>
> Thanks in advance,
>
> Matt
>
> -- 
> Matthew S. Cramer <mscramer at armstrong.com>          Office: 717-396-5032
> Infrastructure Security Analyst                     Fax:    717-396-5590
> Armstrong World Industries, Inc.                    Cell:   717-917-7099
>
> _______________________________________________
> MIMEDefang mailing list
> MIMEDefang at lists.roaringpenguin.com
> http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
>