Even more reliable solution would be to use a monitoring tool like [Monit][1] or [god][2] that automatically detects that a process is dead and restarts it whether it happened during reboot or not. You might think that's an overkill but actually the configuration of Monit is a simple as check process foo with pidfile /var/run/foo.pid start program "foo_start" stop program "foo_stop" if 5 restarts within 5 cycles then timeout Of course the assumption is that the administrator is willing to install the monitoring tool for you so it could be a chicken and egg problem :) [1]: http://mmonit.com/monit/ [2]: http://god.rubyforge.org/