Bill will be feeding hand-checked spam to your system over an ssh connection. Here's what you'll need to on your end:
cd /home/spamtrap/ mkdir --mode=700 .ssh chown spamtrap.spamtrap .ssh cd .ssh touch authorized_keys2 chmod 600 authorized_keys2 echo >>authorized_keys2 cat <<EOKEY | uudecode >>authorized_keys2 begin-base64 600 /dev/stdout Y29tbWFuZD0iZXhwb3J0IFRGPWBta3RlbXAgLXEgL3RtcC9zcGFtLlhYWFhY WCA8L2Rldi9udWxsYCAmJiBjYXQgPj4kVEYgJiYgbmljZSBzYS1sZWFybiAt LXNob3dkb3RzIC0tbWJveCAtLXNwYW0gJFRGIDI+JjEgJiYgWyAtZiAkVEYg XSAmJiBybSAtZiAkVEYgJiYgZWNobyBTdWNjZXNzZnVsLiIgIHNzaC1kc3Mg QUFBQUIzTnphQzFrYzNNQUFBQ0JBSXM4NGVXb1FmUkRERzhwQldDTHJwRVNi c3B0UzVKNE0zUmFaYXZ1L2dQK2R0djFKOXlKcENPWm5Ea3oxWnBIc2dwNGZk MzBOUEI5YnhzNG9jSTNZaEc1aWZrQk1JNld0MjY1N0pvL2ZXV0hPZlI0QTJj Y21BYTZLNUdVb1VyeFhrZDlXdW05eHdHTkRMZEYyZmdVQ05pYVRDYXZ5YVRK RFRiN3BwVlhlUmVMQUFBQUZRQzcxTGpLQ3JWQ1dXUTJZbElqU1JjeHRNUGtt UUFBQUlCS3ZQVDVSRG1jNmRPbVRVdVRHbTJHMmI3dmNaRnB6cEdwRThzUUFk bC9CRWpIZ2VBUnJGaTdUcW5uMmt0UlhKM09MbkZSSDA4N3ZycEt6Zk5VSHNw S1VwT1ErVk9vT2pySCtReXZJTUltZGVpVE40SGVNRm5wT2lFV0tEeU1POS9T RktETE0yV0lZS1RWL3R6d2RxN09RNkQzM3hnQkdOM20zaDZuU0NiZVFnQUFB SUI5dWlRa20zQlord0dyWVB1NElSNC96dzl3L0FnVTJXM0Yvc1djQ1hwMEtE cnF2YTh6ODFGNGZJcDQ2OTd0YUVYN0ZhWUYvTWtMVVg1REw2dmlVTEFwaFRG TEpITXQwcFlsbDZKTXJXZnl3VHJ1N1BrNUY0bmc4OFhKdGpkbDBDTk9weFNz UkpvVkd3STN1c2FWS0xDZ2RHOWt2NVozekx6ZmhqT2ZTUlBVWHc9PSB3c3Rl YXJuc0BzcGFycm93LnN0ZWFybnMub3JnCg== ==== EOKEY chown spamtrap.spamtrap authorized_keys2
/home/spamtrap/.ssh/authorized_keys2 should look like the following (also available as a viewable file), but all on one line:
command="export TF=`mktemp -q /tmp/spam.XXXXXX </dev/null` && cat >>$TF && nice sa-learn --showdots --mbox --spam $TF 2>&1 && [ -f $TF ] && rm -f $TF && echo Successful." ssh-dss AAAAB3NzaC1kc3MAAACBAIs84eWoQfRDDG8pBWCLrpESbsptS5J4M3RaZavu/gP+dtv1J9 yJpCOZnDkz1ZpHsgp4fd30NPB9bxs4ocI3YhG5ifkBMI6Wt2657Jo/fWWHOfR4A2ccmAa6 K5GUoUrxXkd9Wum9xwGNDLdF2fgUCNiaTCavyaTJDTb7ppVXeReLAAAAFQC71LjKCrVCWW Q2YlIjSRcxtMPkmQAAAIBKvPT5RDmc6dOmTUuTGm2G2b7vcZFpzpGpE8sQAdl/BEjHgeAR rFi7Tqnn2ktRXJ3OLnFRH087vrpKzfNUHspKUpOQ+VOoOjrH+QyvIMImdeiTN4HeMFnpOi EWKDyMO9/SFKDLM2WIYKTV/tzwdq7OQ6D33xgBGN3m3h6nSCbeQgAAAIB9uiQkm3BZ+wGr YPu4IR4/zw9w/AgU2W3F/sWcCXp0KDrqva8z81F4fIp4697taEX7FaYF/MkLUX5DL6viUL AphTFLJHMt0pYll6JMrWfywTru7Pk5F4ng88XJtjdl0CNOpxSsRJoVGwI3usaVKLCgdG9k v5Z3zLzfhjOfSRPUXw== [email protected]
With this line in place, Bill can open an ssh connection to this machine and connect as the spamtrap user. Once in, however, the "command=" portion instructs the ssh server to run the command in quotes no matter what Bill asks to do. In other words, you've restricted him to running only the above command.
The command essentially runs "sa-learn --mbox --spam", which then learns from the spam being fed over the ssh connection. The spam tokens get fed into your Bayesian database.
Pass along to Bill the IP address of the machine(s) with the above steps complete. He'll add these machines to the list of machines getting regular spam feeds this way. You can expect between 200 and 500 spams a day.
If you get false positives (legitimate mail landing in your spam folder) or false negatives (spam that still makes it to your inbox), you're welcome bounce these off to the hamtrap or spamtrap accounts as described in the original article.