Matthew T. O'Connor wrote:
Dominic Amann wrote:
I have just configured a postfix -> spamassassin -> anomy/sanitizer
setup myself. It seems to be working very well (only up since 21st
Dec). I have been able to turn _off_ my Mozilla junk mail controls,
and use message filtering (based on X-Spam-Status begins with YES) to
push out the junk (into a place where I can review it).
I am now considering setting up a global account for undetected spam
and misdetected ham in my dbmail. Then I can use fetchmail to run it
through spamassassin and train the bayesian database ( + report to
third party database). My users can use drag and drop to send spam
into the global account, and forward to the ham account to register
false positives. If anyone has a better methodology to train
spamassassin in a global installation let me know.
If you get this working well, please forward the details to the list,
I would very much like to set this up.
Matt
I have placed the filters on two domains which have about 10 e-mail
accounts total for 1-2000 e-mails per day total. At this point, my CPU
load is about 15% in perl, which is spamassassin + sanitizer in no
particular proportion. In a few more days, I will put the filter onto
more domain addresses, and monitor the load at that point.
--
Dominic Amann, Linux Based Solutions Ltd., <http://www.lbs.ca/>
3020 Keele Street, Downsview, ON M3M 2H3
Tel: (416) 270-4587
