might want to try this quickly by doing a
chown spamassassinuser:spamassassinuser /etc/mail/spamassassin/bayes chown spamassassinuser:spamassassinuser /etc/mail/spamassassin/bayes/*
Shane Allen wrote:
========= Config ========= required_hits 7 rewrite_subject 0 report_safe 0 use_terse_report 1 use_dcc 0 use_pyzor 0 use_razor1 0 use_razor2 0 spam_level_char + use_bayes 1 bayes_path /etc/mail/spamassassin/bayes bayes_use_hapaxes 1 bayes_expiry_min_db_size 2000000 bayes_expiry_scan_count 5000 auto_learn 1 auto_learn_threshold_nonspam -2.0 auto_learn_threshold_spam 15.0 ==========================
I have used sa-learn to train the bayesian scanner with approximately
3500 ham and 3000 spam messages.
[EMAIL PROTECTED]:/etc/mail/spamassassin# cat bayes_msgcount |wc -c 2113
So, first, those numbers don't seem to jive. I verified after running sa-learn on my corpuses that it displayed 'Learned from #### messages.', and in each case, it learned from a number of messages appropriate to the corpus that I fed it.
I seem to remember reading in the source or in the pod docs or some such that it used that file as a counter to know when to lock and perform certain operations, so I'm not too worried about that...
However, after training it, I still do not see that it is using bayes rules, nor do I see that it is auto-learning from messages that are being scanned. I receive a message that scores -6.4, and the output of the `cat|wc -c` above does not change. Nor, I believe, do the sizes of the bayes_toks or bayes_seen dbs.
To forestall questions and invite suggestions, the reason we set min_db_size to 2000000 is because disk space is cheap and we did not think that 5MB/100000 rows would be sufficient for a globally used db. Additionally, we are unsure of the impact of having a large db, and it is a small matter to change the min_db_size and then run an expire if it turns out that a larger database slows things down. I am very open to suggestions here.
So... anyone have any ideas why I'm not seeing BAYES_?? rules? An example message sent after the training is below:
==== Msg ====
Return-Path: [EMAIL PROTECTED]
Received: from mail [63.251.38.140] by localhost with POP3 (fetchmail-6.2.0)
for [EMAIL PROTECTED] (single-drop); Wed, 26 Mar 2003 23:35:21 -0600 (CST)
Received: by onemail.net (CommuniGate Pro PIPE 3.5.9)
with PIPE id 15721593; Wed, 26 Mar 2003 23:34:54 -0600
Received: from [66.138.218.169] (HELO escaflowne)
by onemail.net (CommuniGate Pro SMTP 3.5.9)
with ESMTP id 15721353 for [EMAIL PROTECTED]; Wed, 26 Mar 2003 23:34:41 -0600
Received: from shane by escaflowne with local (Exim 3.36 #1 (Debian))
id 18yQ2D-0008Tq-00
for <[EMAIL PROTECTED]>; Wed, 26 Mar 2003 23:34:41 -0600
Date: Wed, 26 Mar 2003 23:34:40 -0600
From: Shane Allen <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
Subject: testing bayes stuff
Message-ID: <[EMAIL PROTECTED]>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.5.4i
X-Spam-Status: No, hits=-6.4 required=7.0
tests=USER_AGENT_MUTT
version=2.52
X-Spam-Level:
X-Spam-Checker-Version: SpamAssassin 2.52 (1.174.2.8-2003-03-24-exp)
X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.11.1.3
Hopefully I will see a bayes_?? rule above, and I will be able to cancel the email to satalk. =============
-- Shane Allen <[EMAIL PROTECTED]>
------------------------------------------------------- This SF.net email is sponsored by: The Definitive IT and Networking Event. Be There! NetWorld+Interop Las Vegas 2003 -- Register today! http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
------------------------------------------------------- This SF.net email is sponsored by: The Definitive IT and Networking Event. Be There! NetWorld+Interop Las Vegas 2003 -- Register today! http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
