If you can't detect failures, you are unable to recover from them. That's an obvious principle you need.

— Joe Armstrong, Faults, Scaling and Erlang concurrency

Topics: debugging, distributed systems