1. If you don't already have a submission account, send a request to
submit@spamassassin.org and ask for a GA mass-check submission account.
You will receive your username and password via email, it will only be
good for GA mass-check submissions. If you're interested in the nightly
submissions, please see the CORPUS_SUBMIT_NIGHTLY document.
2. Get the latest version of SA following the instructions in the GA
mass-check announcement email.
3. Now cd to the "masses" directory in the checked-out CVS code tree.
4. Read README to gain understanding of what mass-check does.
5. Run mass-check against your ham mail archive.
6. sort -rn +1 ham.log | head -20
7. Check each of those 20 messages by hand to make sure they're not spam that
slipped through, or a forward of a spam message.
8. Repeat #6 until the top 20 are "clean"
9. Repeat steps 4-7 for your spam archive until they are "clean"
(except you do sort -n +1 spam.log to look for low scoring spam)
10. Run a mass-check for ham and spam together (one mass-check run)
11. rename ham.log and spam.log to the appropriate filenames. ** see note below **
12. rsync -CPcvzb ham-yourname.log spam-yourname.log username@rsync.spamassassin.org::submit
Thanks for your help!
Note: Depending on what type of mass-check you've run, the name of
the file you need to upload may be different. The different types of
mass-check are combinations of with/without Bayes, and with/without
Net rules. The resulting filenames are:
Set 0 ham-nobayes-nonet-username.log spam-nobayes-nonet-username.log
Set 1 ham-nobayes-net-username.log spam-nobayes-net-username.log
Set 2 ham-bayes-nonet-username.log spam-bayes-nonet-username.log
Set 3 ham-bayes-net-username.log spam-bayes-net-username.log
For GA mass-check runs, there will be 3 announcements for people to run
sets 1-3 (we can get the set 0 results by removing the net results from
set 1 ...)