I am an experienced developer on Windows, but not too experienced on Linux. I am playing with memcached on CentOS, and I have achieved that it starts on boot by using chkconfig. However when I kill the process (by sudo kill PID) it never starts again. I know memcached is the most reliable application ever, but conceptually I'd still like to know that if it dies it can get a second chance.
In Windows in services there is a configuration for recovery where you can specify what you want to do if the process hosting the service dies (hence this concept is useful at least on Windows), but I did not find such a thing with chkconfig. I wonder if the Linux service management has something similar that watches a service and restarts it if it dies.