Bayes not working?

Peter Lemieux mailscanner at replies.cyways.com
Thu Sep 14 17:41:03 UTC 2017


I grepped my logs for "autolearn=spam".  At the bottom, there were only 
three messages with Spam Assassin scores between 6.0 and 6.5.  All the rest 
were higher, and the vast majority had double-digit scores.  If you have 
"Log Spam" enabled, grep your logs for 'is spam'.  What kinds of scores do 
you see?

My default threshold for tagging a message as {Spam?} is 4.0 ("Required 
SpamAssassin Score" in MailScanner.conf).  I quarantine messages with scores 
above 10.0 ("High SpamAssassin Score").  Generally speaking 5.0 is 
considered the smallest minimum that avoids false positives, but my 
experience suggests that's too permissive.

Peter


On 09/14/2017 05:22 AM, Mark Meelhuysen wrote:
> OK, so it is possible that there are no valid e-mails yet. I will check the 
> score of some spam items.
> 
> How about the ORGNAME in the config file? Do you know anything about that? 
> Changed it to check what happens. I see no difference. I thought it might be 
> passing items through because of this setting?
> 
> *Van:*MailScanner 
> [mailto:mailscanner-bounces+mark=meelhuysen.com at lists.mailscanner.info] 
> *Namens *David While
> *V**erzonden:*donderdag 14 september 2017 10:38
> *Aan:* MailScanner Discussion <mailscanner at lists.mailscanner.info>
> *Onderwerp:* Re: Bayes not working?
> 
> Ok Now I am at a computer!
> 
> Default threshold for spam is 12. This can be changed with a config setting 
> but the minimum value is 6.
> 
>  From the spamassassin web page:
> 
> 
>   USER OPTIONS
> 
> The following configuration settings are used to control auto-learning:
> 
> *bayes_auto_learn_threshold_nonspam n.nn (default: 0.1)*
> 
> The score threshold below which a mail has to score, to be fed into 
> SpamAssassin's learning systems automatically as a non-spam message.
> 
> *bayes_auto_learn_threshold_spam n.nn (default: 12.0)*
> 
> The score threshold above which a mail has to score, to be fed into 
> SpamAssassin's learning systems automatically as a spam message.
> 
> Note: SpamAssassin requires at least 3 points from the header, and 3 points 
> from the body to auto-learn as spam. Therefore, the minimum working value 
> for this option is 6.
> 
> On 14/09/2017 09:29, David While wrote:
> 
>     Not all spam triggers the autolearn. I can't remember what the number is
>     but the spamassassin score has to be above a certain number to trigger
>     the autolearn.
> 
>     Sent from BlueMail <http://www.bluemail.me/r?b=10455>
> 
>     On 14 Sep 2017, at 09:24, Mark Meelhuysen <mark at meelhuysen.com
>     <mailto:mark at meelhuysen.com>> wrote:
> 
>         And i was looking but unable to find Bates!! JNo, just kidding..
> 
>         In my config auto learn is enabled. It is by default.
> 
>         I concluded bayes not working as I saw indeed that there are 0 spams
>         in the DB.
> 
>         The config is running for a couple of days. It detected some spam,
>         does that mean that when the system detected 10 e-mails as spam that
>         my Bayes DB should contain 10 spam emails?
> 
>         *Van:*MailScanner
>         [mailto:mailscanner-bounces+mark=meelhuysen.com at lists.mailscanner.info]
>         *N**amens *David While
>         *Verzonden:* donderdag 14 september 2017 10:20
>         *Aan:* MailScanner Discussion <mailscanner at lists.mailscanner.info>
>         <mailto:mailscanner at lists.mailscanner.info>
>         *Onderwerp:* Re: Bayes not working?
> 
>         Damn autocorrect! Should obviously be Bayes not Bates!!
> 
>         Sent from BlueMail <http://www.bluemail.me/r?b=10455>
> 
>         On 14 Sep 2017, at 09:18, David While <dwhile at while.org.uk
>         <mailto:dwhile at while.org.uk>> wrote:
> 
>             There appears to be 0 spam emails in the Bates database. Check
>             that you have Bates set to autolearn otherwise you will not get
>             any.
> 
>             Sent from BlueMail <http://www.bluemail.me/r?b=10455>
> 
>             On 14 Sep 2017, at 09:06, Mark Meelhuysen < mark at meelhuysen.com
>             <mailto:mark at meelhuysen.com>> wrote:
> 
>                 Hi All,
> 
>                 I noticed that my Bayes DB is not building. My spamassassin
>                 –D –lint output for bayes:
> 
>                 Sep 14 10:02:06.389 [50764] dbg: bayes: tie-ing to DB file
>                 R/O /etc/MailScanner/bayes/bayes_toks
> 
>                 Sep 14 10:02:06.389 [50764] dbg: bayes: tie-ing to DB file
>                 R/O /etc/MailScanner/bayes/bayes_seen
> 
>                 Sep 14 10:02:06.389 [50764] dbg: bayes: found bayes db version 3
> 
>                 Sep 14 10:02:06.390 [50764] dbg: bayes: DB journal sync:
>                 last sync: 0
> 
>                 Sep 14 10:02:06.390 [50764] dbg: bayes: not available for
>                 scanning, only 0 spam(s) in bayes DB < 200
> 
>                 Sep 14 10:02:06.390 [50764] dbg: bayes: untie-ing
> 
>                 Any idea’s where to start looking? Bayes and spamassassin
>                 are not generating errors in MailScanner –lint and
>                 spamassassin –D --lint
> 
>                 Met vriendelijke groet,
> 
>                 Mark Meelhuysen
> 
> 
>                 -- 
>                 This message has been scanned for viruses and
>                 dangerous content by *MailScanner*
>                 <http://www.mailscanner.info/>, and is
>                 believed to be clean.
> 
>                        
> 
>                  
> 
> 
>         -- 
>         This message has been scanned for viruses and
>         dangerous content by *MailScanner* <http://www.mailscanner.info/>,
>         and is
>         believed to be clean.
> 
> 
>         -- 
>         This message has been scanned for viruses and
>         dangerous content by *MailScanner* <http://www.mailscanner.info/>,
>         and is
>         believed to be clean.
> 
> 
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
> believed to be clean.
> 
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
> believed to be clean.
> 
> 
> 
> 


More information about the MailScanner mailing list