Hi all, I read all the docs but still have some questions.
At the moment I'm using spamassassin 2.60 together with sendmail and spamass-milter. My /etc/mail/spamassassin/local.cf uses global settings for all users, also for global bayes. Users (about 600) aren't shell-users (/etc/passwd) but have lokal dirs for vpopmail-like system. Now I consider to have a) per user definitions in sql or like vpopmail options in users directories. b) a possibility for our users to claim a mail as spam and feeding it back to bayes (and with spamassassin -r also to razor/pyzor) Concerning a): - as long as there is no special user pref in sql, SA uses @GLOBAL. What happens if sql server is'n running? What preferences then will take effect? How can I set setting for all users of a domain? Can I write in SQL a setting for user '@some.domain.de' ? Is thsi overwritte by per user definition in SQL? - How can I realize a per user bayes, if also exists an sql entry for this user? SA should use global bayes for all others. Is this better done by a vpopmail like setting in loakl users directories? What will then used as fall back setting, if there is no user pref? Can I also have per user bayes and a fallback global one? Conc. b): Mail is accepted by sendmail, then sent by uucpdom mailer to our special software, then maisl will be stored in mail folders. So there are some changes in headers from all this sending and forwarding. If users now forward a mail as spam to a special lokal user, do i have to delete/coorect these headers for bayes filter? What about arzor/pyzor? I noticed SA to delete own headers when reporting a mail as spam, but safe reports aren't set back to their original appearings like when mail comes in. So, is it best to NOT report an already recognized spam? If I want to have my users use their very own bayes database, how can i setup a mail address for reporting spam and learning 'their' bayes system? (Well, I don't trust my users to correctly identify spam. I fear them reporting old mailing lists, they just don't want to get but forgot to unsubscribe. So I don't want a global bayes feedback ;-) .) Would be nice to get some opinions, hints or even URLS to find more about that. thank you Andreas ------------------------------------------------------- This SF.net email is sponsored by: The SF.net Donation Program. Do you like what SourceForge.net is doing for the Open Source Community? Make a contribution, and help us add new features and functionality. Click here: http://sourceforge.net/donate/ _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk