spamassassin-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Jones <djo...@ena.com>
Subject Re: Question about BAYES_999
Date Tue, 02 Jan 2018 00:59:14 GMT
On 01/01/2018 06:52 PM, David Jones wrote:
> On 01/01/2018 06:47 PM, Reindl Harald wrote:
>>
>>
>> Am 02.01.2018 um 01:18 schrieb David Jones:
>>> I just had a spam message hit BAYES_999 but not BAYES_99.  Based on 
>>> BAYES_999 default score of 0.2, I thought that it was always supposed 
>>> to complement the BAYES_99 rule and both would trigger when BAYES_999 
>>> hit.
>>>
>>> https://pastebin.com/QsVgXwdC
>>>
>>> If they are independent, then it would seem logical to bump up the 
>>> default score higher than BAYES_99
>>
>> never ever seen that and since bayes is based on a number between 0 
>> and 1 this should be technically impossible at all
>>
>> with BAYES_00 that message has [score: 0.0003]
>>
> 
> I checked my logs and I am seeing both together when BAYES_999 hits 
> except for a few times.  Is this a bug?  Should I open a bug issue?  I 
> am not sure how to reproduce the problem unless others also see the same 
> thing with that message.
> 

Sorry.  Not thinking clearly.  Others would have to have the same Bayes 
DB to get that message to do the same thing.  I was able to reproduce 
the same results on another SA platform running MailScanner using the 
same Bayes DB in redis.

If others could check their mail logs to see if they are hitting 
BAYES_999 without BAYES_99 on the same message, please let me know.

-- 
David Jones

Mime
View raw message