Re: [AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-08 Thread mouss
Christopher Jack a écrit : > Hi there, > > I've been looking into this very problem over the last few days and wondered > how you got on with your solution as I'm in the same boat as you. That is > that my amavisd-new box and my web host are different machines. > > How did you get on with the

RE: [AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-07 Thread Christopher Jack
hristopher. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of BJ Dierkes Sent: 03 February 2006 21:00 To: amavis-user@lists.sourceforge.net Subject: Re: [AMaViS-user] Re: Best way to train statistical DB's ? Felix Schwarz wrote: >>Wow.. hm. that s

[AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-05 Thread mouss
Jakob Curdes a écrit : > >> Try the command line option '--user="[EMAIL PROTECTED]"'. Do you use >> DSPAM with virtual users? >> >> >> > We have many domains on the server and user postfix as MTA. We use the > postfix virtual table to map mail addresses to system users, but all > mail users (i.e

Re: [AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-03 Thread BJ Dierkes
Felix Schwarz wrote: Wow.. hm. that should work for my one config. Howerver, Any ideas if mail.mydomain.com and www.mydomain.com are two different servers? I.e. Horde is on a different box than amavisd-new is. Somehow you have to get the data from the web host to the amavisd-new h

Re: [AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-03 Thread Jakob Curdes
Try the command line option '--user="[EMAIL PROTECTED]"'. Do you use DSPAM with virtual users? We have many domains on the server and user postfix as MTA. We use the postfix virtual table to map mail addresses to system users, but all mail users (i.e. pop/imap accounts) are system accounts

[AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-03 Thread Felix Schwarz
BJ Dierkes schrieb: > Felix Schwarz wrote: >>That depends on what filter you are using. SpamAssassin's bayes db >>itself can be used by different users. amavisd-new does currently only >>use one user. If you want to be able to report spam by your webmailer >>you should make sure that both are usin

[AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-03 Thread Felix Schwarz
Hallo Jakob, Jakob Curdes schrieb: > I figured out a way to train dspam via mail forwarding but got stuck > in the user jungle - dspam wants to train the mail as user "nobody" > and dies because it cannot create the lock files for this user. Try the command line option '--user="[EMAIL PROTECTED]"

[AMaViS-user] Re: Best way to train statistical DB's ?

2006-02-03 Thread Felix Schwarz
Hi BJ, BJ Dierkes schrieb: > I am using Horde Webmail. I have configured it so that when you > click "report as spam" it runs sa-learn against the email. Now, this > is being run as www-data (debian) I assume and therefore my question > is this: Are the statistics processed into a global database