Bayes issue

[Shawn Iverson]

That looks correct.

On 7/23/20 4:54 PM, Miron Jajtić wrote:
> It's symlinked like this:
>
> ===========================
> [root at smtp ~]# stat /etc/mail/spamassassin/MailScanner.cf
>    File: ‘/etc/mail/spamassassin/MailScanner.cf’ -> ‘/etc/MailScanner/spamassassin.conf’
>    Size: 34            Blocks: 0          IO Block: 4096   symbolic link
> Device: fd00h/64768d    Inode: 1704809     Links: 1
> Access: (0777/lrwxrwxrwx)  Uid: (    0/    root)   Gid: (    0/    root)
> Access: 2020-07-21 21:45:40.558652022 +0200
> Modify: 2020-07-21 21:45:40.558652022 +0200
> Change: 2020-07-21 21:45:40.558652022 +0200
>   Birth: -
>
> [root at smtp ~]# stat /etc/MailScanner/spamassassin.conf
>    File: ‘/etc/MailScanner/spamassassin.conf’
>    Size: 14619         Blocks: 32         IO Block: 4096   regular file
> Device: fd00h/64768d    Inode: 1704870     Links: 1
> Access: (0644/-rw-r--r--)  Uid: (    0/    root)   Gid: (    0/    root)
> Access: 2020-07-22 00:46:38.602638839 +0200
> Modify: 2020-07-22 00:46:38.602638839 +0200
> Change: 2020-07-22 00:46:38.611638569 +0200
>   Birth: -
> ===========================
> Is this somehow wrong?
>
>
> Best regards,
> Miron
>
> On 23. 07. 2020. 22:42, Shawn Iverson via MailScanner wrote:
>>
>> Verify that your /etc/mailscanner/spamassassin.conf is properly 
>> symlinked to your spamassassin config directory.
>>
>> On 7/22/20 12:08 PM, Miron Jajtić wrote:
>>> Hello to everyone!
>>>
>>>
>>> I'm trying to solve issue with Bayesian filtering.
>>>
>>> For some reason one message is constantly marked as spam due to this:
>>>
>>> ==================================
>>>
>>> Spam Report:
>>> Score    Matching Rule    Description
>>> 15.00    BAYES_99    Bayes spam probability is 99 to 100%
>>> 15.00    BAYES_999    Bayes spam probability is 99.9 to 100%
>>>
>>> ==================================
>>>
>>>
>>> No matter does I disable or enable bayes inside of
>>> /etc/MailScanner/spamassassin.conf or change bump score to negative, etc.
>>>
>>>
>>> ==================================
>>>
>>> [root at smtp MailScanner]# grep -i bayes /etc/MailScanner/spamassassin.conf
>>>
>>> # =============== Bayesian Filtering ===============
>>> # By default, the Bayesian engine is used. This is a real CPU hog
>>> # use_bayes 0
>>> use_bayes 1
>>> bayes_auto_expire 1
>>> bayes_store_module              Mail::SpamAssassin::BayesStore::Redis
>>> bayes_sql_dsn                   server=xx.xx.xx.xx:6379;database=0
>>> bayes_token_ttl 21d
>>> bayes_seen_ttl   12d
>>> bayes_auto_learn 1
>>> bayes_auto_learn_threshold_nonspam    -2.0
>>> bayes_auto_learn_threshold_spam       5.0
>>> # Bump up SA score if it matches something in Bayes
>>> score BAYES_00 -5.0
>>> score BAYES_05 -1.5
>>> score BAYES_50 1.5
>>> score BAYES_60 2.0
>>> score BAYES_80 6.0
>>> score BAYES_95 9.0
>>> #score BAYES_99 0
>>> #score BAYES_999 0
>>> ==================================
>>>
>>>
>>> What's interesting is when I test the same message directly with
>>> Spamassassin and using above config file, then bayes rules are not
>>> activated.
>>>
>>> After two days of debugging, I'm out of idea where to check why is this
>>> message marked with 30 spam points when is processed via MailScanner.
>>>
>>> I would be really grateful if someone can point me in the right
>>> direction how to resolve this issue
>>>
>>>
>>> ==============
>>>
>>> MailWatch: 1.2.15
>>> OS: CentOS 7
>>> MailScanner: 5.3.3
>>> SpamAssassin: 3.4.4
>>>
>>> ==============
>>>
>>>
>>> Best regards,
>>> Miron
>>>
>>>
>>>
>>>
>> -- 
>>
>> Shawn Iverson
>> shawniverson at summitgrid.com
>>
>
>
>
-- 

Shawn Iverson
shawniverson at summitgrid.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mailscanner.info/pipermail/mailscanner/attachments/20200723/714466c3/attachment.html>