Most email users are blissfully unaware of the magnitude of the spam problem. Major email providers do a pretty good job of filtering, so you might see little or no spam in your inbox. Even the stuff that gets sent to your “spam folder” is only a tiny fraction of the spam that gets sent to you. That hides the magnitude of a really big problem, and one you pay for even if you don’t realize it.

The Sacdoc email server handles email for Sacramento Medical Oasis, Inc. along with the risley.net, twoprops.org, and a handful of other domains. My email is the busiest account, and I average about two hundred legitimate emails a day. Along with that, the server gets hit with about ten thousand attempts to send spam to my address. Ten thousand, every day.

To handle that mass of email slime, I run a free and open-source filter on my mail server called Apache Spamassassin. It’s an impressive bit of code, and is used by many email providers large and small.

I was reminded of its importance today when there was an update to Spamassassin (to 3.4.2, the first update in 3½ years!). After I installed the update, everything appeared to be running correctly but massive quantities of jellied meat byproducts were appearing in my inbox. Actually, most were caught by secondary filters so ended up in my spam folder, but they were arriving faster than I could delete them.

If you’re a mail server admin and this happens to you, it turns out you need to update Spamassassin’s rule set after the update:
/usr/local/bin/sa-update --debug channel && /etc/init.d/spampd restart
Since Spamassassin updates happen so infrequently, it’s easy to forget about that and the answer is hard to find with a search of the web. (With all respect to the Spamassassin development team, who is giving us a fantastic bit of software absolutely for free, it would be nice if this quirk were mentioned in the release notes or if the install process itself triggered a rule update.)

Readers might be interested to know that processing spam uses more computing resources than any other function that the Sacdoc/risleynet/Medical Oasis server clusters perform. The amount of computing resources (and power and hardware and administration time) consumed by the spam problem worldwide must be astronomical.