Hi I have an aging PC running exim/spamassassin, which is also my internet gateway and a load of other stuff. Machine goes down and the house is without an internet connection.
So in moving away from that, I decided to use a small Raspberry Pi as a mail server (and some other stuff) for the home. 4 separate mail address (2 of them are mine, personal and business). The others are for my wife and daughter - both NOT computer savy. We have about 8G of mail stored - I have a 32G thumb drive on the PI set up as lvm with /var and /var/mail stores as logical volumes on it. This is great because I have already shown migrating that from one drive to another without taking the the mail serving part offline is easy. With a lot of learning, I now have a postfix -> dovecot setup and 4 milters for before queue filtering. These are milter-greylist (only used for spf checking), dkim-milter (dkim checking - maybe in future dkim signing, although for now my domain name host doesn't support dkim key records), clamav-milter -> clamav-daemon for virus scanning and dspam-milter->dpam for spam checking. Dovecot has a sieve setup, with (for now - more to come) a default sieve that handles postfix sender_bcc of outgoing mail back to a user with a +saveout extension to move mail to a Sent mail store. Dovecot is set up to use standard Mailstore's but there are no user accounts on the machine, everything is treated like a virtual mailbox - and Dovecot is also Postfix Authentication mechanism via a small sqlite database. As of now, all seem to be working - at least I have been trying to break it for a while, although its was only this evening that I got the dspam-milter/dpam daemon in place and working. Now I need to ensure that dspam can train itself but I don't really understand it enough to get move away from its use of mail addresses - not entirely convenient. I thought I would ask on this list for more advice. Both my wife and I use Thunderbird as our main mail client and it and my wife (and myself) seems quite adapt at moving mail they think is Junk into the junk folder (a special use folder defined in Dovecot to be for \Junk). So from a user perspective, by far the most natural approach for training for spam would be to use something like a cron job to regularly scan each users junk Maildir using the dspam_train command line tool in a script. Is this an accepted approach.? What should I use as ham? Do I need as much ham as spam? I would very much appreciate some advice on the matter. One further question - I don't really understand python, so it took my a while to get dpam-milter working. I run it from debian start-stop-daemon, but if I try and run it other than root (I was hoping to run it as dpam) it disappears silently without trace. Is there any likely causes for this, that I can fix easily. Thanks Alan ------------------------------------------------------------------------------ Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce. With Perforce, you get hassle-free workflows. Merge that actually works. Faster operations. Version large binaries. Built-in WAN optimization and the freedom to use Git, Perforce or both. Make the move to Perforce. http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk _______________________________________________ Dspam-user mailing list Dspam-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspam-user
