> > > LOL, it is official, I am taking my own advice. I am upgrading my live > > server from 2.4x! And I think the experiment was a complete success!!!! > > > > I never upgraded live server from 2.4x because I wanted to see how good I > > could get that system working without Bayes and net tests. 2.4x ran > > perfectly the whole time! Caught about 99%. This was due to BigEvil and > > SARE. (Also denying at the MTA level with standard RBLs.) > > > > I wanted to prove to myself that Bayes wasn't needed, and I did. I intend > > to > > let anyone know that complaining about SA not catching all, is wrong! > > Proper > > administration of SA is the key. If a 2 year old version (might be older > > then that!) can block 99%, then it ain't the software ;) > > > > The main reason I'm updating is SURBL support, and being on the same page > > as > > the rest of the SARE ninjas. (They were making fun of my old sword!) > > > > The devs should be proud that their older version still kicks butt. And > > when > > you read all the dooms day articles on spam by the media, sit back and > > chuckle with me. Think to yourself, "Hell, the solution was made many > > years > > ago, you just need to use it!" > > > > Well....on to reconfiguring the server! If I don't answer peoples' email > > in > > the next few days....something went horribly wrong ;) > > > > --Chris (Hmmm.... Perl version 5.00x running.....this is gonna be fun!) > > > FWIW, I use SA 3.0.0 and spamass-milter on a relay server that hands > messages through the firewall to the "real" mail server. As such, we can't > use Bayes (or perhaps I should say "I am not good enough yet to figure out > how to use Bayes in this environment"). It still catches almost all spam.
Well in SA v3.0 when Bayes score 99-100% it only adds 1.9 to the score as opposed to the 5.4 it added in v2.6. Sure you can override the score, but what I'm trying to say is it seems the new SA does not rely that heavily on Bayes any more. Danie
