Help!! URGENT!!

[kfliong]

bayes_auto_learn option is not in spam.assassin.prefs.conf. So, I just add 
that option into it after use_bayes?

As for spams, yes, we do get like 90% of spams in our mails. And it's 
really getting out of control. I have suspected people running spam from 
our dedicated server and have checked but can't find anything. It really is 
a nuisance. What else can I do to solve this spamming problem? There is no 
one that can help and ev1servers.net is no good either.

We don't mind paying for the service is someone could kill off all these 
spammers. Would not mind to give access to the person on our server.

Thanks in advance.

At 07:15 PM 3/30/2004, you wrote:

> >-----Message d'origine-----
> >De : kfliong [mailto:kfliong at WOFS.COM]
> >Envoyé : 30 mars, 2004 02:32
> >À : MAILSCANNER at JISCMAIL.AC.UK
> >Objet : Re: Help!! URGENT!!
> >
> >
> >Yup, I have checked the mails and most of them became spam
> >because BAYES_99
> >have a value of 5.90. That itself already make the email a
> >spam. Default
> >spam score is 5.00.
> >
> >Anyway, does turning "use_bayes 1" meaning that spamassassin will
> >automatically learn all the emails that go through
> >mailscanner?
>
>set
>
>bayes_auto_learn 0
>
>This will turn off auto-learning.
>
>I think you should investigate how come you get so many spams.  Sure you 
>don't have Infinite-Monkeys in MailScanner.conf?
>
>Ugo
>
> >How can I
> >switch on bayes and only make it learn from messages that I fed it?
> >
> >At 12:26 PM 3/30/2004, you wrote:
> >
> >>On Tue, 30 Mar 2004 01:19 pm, Stephen Swaney wrote:
> >> > Hopefully you don't have
> >> >
> >> > Infinite-Monkeys Anywhere in your MailScanner.conf files.
> >> >
> >> > Steve
> >> >
> >> > Stephen Swaney
> >>
> >>This is an honest question: what has that got to do with
> >SpamAssassin's
> >>Bayesian engine?  The original message was about rapidly growing bayes
> >>database and BAYES_99 rules popping up in every message.
> >Does the bayesian
> >>engine hook into RBL's somehow?
> >>
> >>I know about Razor and Pyzor providing "confidence" levels
> >for spam since SA
> >>2.61 etc....but that's not anything to do with th *local* Bayesian
> >>classifier.
> >>
> >>Just curious - have I missed something?
> >>
> >>James
> >>--
> >>Fortune cookies says:
> >>"I'd love to go out with you, but it's my parakeet's bowling night."
> >
> >thanks
> >

thanks