Building a Spamtrap with Spamassasin

I am currently trying to build a spamtrap with Spamassasin. The idea: I add
another emailaddress to my website. Every email to that address is piped to “sa-learn –spam”. My problem, sa-learn needs to save the gathered information somewhere, and i am not sure where. It looks like spamc stores the information from autolearning on the spamd-server, but sa-learn stores the information local. I don’t want to run a mailserver on the spamd-machine just for this task. Any ideas?

5 thoughts on “Building a Spamtrap with Spamassasin”

  1. you can tweak the path where autolearning results are stored via “bayes_path /path” in local.cf . In sa-learn you can use this config by telling sa-learn where your config resides via the “-C” Option – for example:
    sa-learn -C /etc/mail/spamassassin –showdots –ham –dir /home/lynix/ham.in/
    hope this helps

  2. btw: try to avoid false positives .. feed more ham than spam.

  3. Thanks! With the –dir option i think i can run sa-learn directly from the alias file, and don’t need to extend my .procmailrc.
    But I still have the problem how to transfer the data to the machine running spamd. Maybe I need to switch from spamc/d to using SA direct on the mailserver

  4. .. why just not copy them an learn them via cronjob?
    on the other hand – SA on the mailserver itself and a script for reporting ham/spam would also be a good option.

  5. Another option is to learn from an mbox� and send the results to your spamd server using the host (-h) option. Requires no modification of your procmailrc only a regualr cronjob if your spamd is available.
    �)I store all incoming UBE/UCE in an single file using procmail.

Comments are closed.