When I was done, I ran the following to test everything:
sudo –u postfix Mailscanner –lint
Then I wrote a script that does an sa-learn every night once all my mail has been moved around. It learns not only what is considered Spam, but also what is considered to be Ham. I ran this script periodically during the first few days by hand to train SpamAssassin with what is already there.
This went fairly smoothly, but I wanted to have better control over user passwords, so I implemented DIGEST-MD5 passwords into Postfix.Admin and donated the patch back to Postfix.Admin. There are still a few issues, but this allows you to store your password in DIGEST-MD5 mode. I have to still work on the ability to submit via DIGEST-MD5, etc.
In addition, I wanted to be able to better control my whitelists in MailScanner (there is a method to do this in MailWatch, but it too is broken). I hope to get that patch uploaded shortly as well.
I now have an easy to monitor email subsystem that tells me that I receive roughly 49.5% of high scoring Spam. I will continue to refine this system and perhaps add in other spam fighting techniques, but for now the system runs quickly and allows easy monitoring and a quick way to manage whitelists, aliases, and virtual domains.
Now, this runs within a well monitored VM, but not within a Cloud just yet.