Re: moderately personalized spam sneaking past my SA. general approaches to fix it?

> On Thu, Nov 21, 2024 at 12:42:58PM -0500, pgnd wrote:
>> , sneaking past my SA protections.  Whether to call it snowshoe, I dunno :-/  But my usually well fed Bayes isn't getting noticeably better with 'em.
>> 	https://gist.github.com/pgnd/5ab934d921939f4c62a4c978a30b9e6f
> 
> Are you even running bayes? Headers do not mention any BAYES_* matches (not even the undecisive BAYES_50) ?
> Is something shortcircuiting Bayes, or is it perhaps not running for some other reason?

Good catch.

Yes, I am, in general. And on most if not all the machines where these are sneaking through.

Although ...

In this example it's off(... 'usually' ...) while I'm figuring out just what the non-Bayes rules are doing.
I use

	bayes_store_module  Mail::SpamAssassin::BayesStore::Redis

and figured it wise to get that out of the loop.
I did see my entries being written during learning, but wasn't seeing significant scoring as a result for these leak-thrus.
I'll put it back eventually.

I'm focused atm on using available rules -- out-of-the-box and KAM.
IF the answer is that the only way to cope with this is WITH Bayes, then that's a learning too.