Is this tutorial still applicable to the current version of Instiki? http://rubyforge.org/docman/view.php/85/107/instiki_service_tutorial.txt If so, why isn''t this approach mentioned here? http://instiki.org/show/Running+as+a+Windows+Service We''ve used the latter approach and are unhappy with it. It starts Instiki as a service, but has no idea if Instiki fails. Last week we implemented logging for our Instiki service by modifying config/environments/production.rb to define ActionController::Base.logger to point at a log file. However, the next time the system rebooted, Instiki didn''t come up, and what''s more, there was no indication that the service had failed. Nobody was notified, and the service status appeared to still be running. Well, it turns out that the path to the log file was to a network drive that wasn''t mounted before the service was started. I''ve since then solved that problem (I hope -- we''ll see on the next reboot) by switching to using a UNC path. The point of this little tale is not about logging to a network drive (which should probably be avoided, for maximum reliability) but that services started using the method documented on the instiki.org wiki can fail without anyone knowing about it! I am guessing this is because the service starter utility does not watch to see if the application dies, but simply "starts and forgets" the application. I am hoping there is a better way, such as the win32-service tutorial pasted above, and that this can be integrated into Instiki instead of being left as an exercise for each administrator to implement on their own. Thanks, Ben