Test Drive System Administration – By Steven Butterworth
Many programmers have at least passing familiarity with Test Driven Development, wherein tests are written along with, or even in advance of, the functional code. In this talk, I will discuss the idea of Test Driven System Administration. Working in conjunction with conventional service and host monitoring, we can improve reliability and security by writing tests and generating reports that confirm our systems are in the state that we think they are. With the right infrastructure in place, we can run those tests periodically or even when specific system state changes. – Are the permissions on that filesystem still consistent with your intended settings? – Does every storage volume have a backup volume? Do the backup files have the same content as the source files? – You installed that latest OpenSSL patch, right? Did you also check that it really mitigated the exploit? Do you check again every night? – You just dropped the firewall to perform some testing. Did you remember to turn it back on? We all make mistakes. Let’s get the system to tell us before they cause problems.