Was: Upgrade fron 4.61.7 to 4.74.13-2

Glenn Steen glenn.steen at gmail.com
Fri Jan 9 20:45:41 GMT 2009

2009/1/9 Kai Schaetzl <maillists at conactive.com>:
> c.on.ca>
> Reply-To: mailscanner at lists.mailscanner.info
> Dave Filchak wrote on Fri, 09 Jan 2009 12:10:06 -0500:
>> So, I need to either change
>> the spam actions for not high spam to striphtml deliver,
> Of course, it's really not a good idea to quarantine low scoring spam.
> What you do is fine-tune your setup and then lower *carefully* the high-
> scoring spam from 10 downwards (I have it set to 6). And the lowering
> coincides with finetuning your detection in SA and your Bayes.

This is good advice from Kai to you Dave.
Also... I guess you've already looked at the sections in the MAQ/Wiki
about getting the most out of SA? Else... that's a good startingpoint.
And as friend Scott says... it might be time to try retrain your
bayes... At least have a look at what scores you get.
Since you use MailWatch, you have a superb tool in the reports page,
for determining exactly what makes the lowscoring spam (just above 5)
end up there.  Apply filters for "Is spam = 1" and "SA score < 6",
then look at the SA rule hits report... and perhaps browse throgh the
details for a few of them, just to get a feel for what's wrong.

> Kai

-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se

More information about the MailScanner mailing list