If you're a regular reader here, or if you've seen me speak at conferences, you know that Security is a big focus of what I do. I've been recently trying out a nice tool for Linux servers that's designed to stop dictionary attacks and worms and such. The tool comes with pre-made configurations for common linux packages like ssh, ftp, and apache web server but it can be easily expanded to include other applications.

Fail2Ban essentially watches log files for entries showing failed password or other such problems, and can be set to add a firewall rule for sites that repeatedly fail. The ban can be short term or long term as well. I've got mine set to block connections from any IP address that fails a credential challenge more than (n) times in (s) seconds and to keep that IP address blocked for (m) minutes. It seems to work brilliantly. Everyone once in a while I get a notice by email that it blocked a site (usually from Russia or China) but otherwise everything on the system proceeds normally.

SMTP is a very common attack vector because there is so much garbage on that port that many sites don't watch their logs well. Any site that requests credentials for sending SMTP offers the promise of a fantastic spam relay point. I've seen more than one Domino server where a developer account (e.g. Test User) with a common password (e.g. password) was enabled for web access. I've seen these server unknowingly become spam servers because some scripted spam server found that common login worked and they were able to send spam as a trusted user of that server. I go on at some length in my security lectures at conferences about the need for good password management rules on Domino's http password feature even if you don't think you're using it, and this is why.

Fail2Ban is a linux package, and won't work in a standard Windows server environment. I've considered writing something similar for that environment though. Who knows.