have you tried turning it off and on again?
expressionIT supportops culture
Support's opening liturgy and its most statistically justified question: a restart clears the accumulated state where most transient faults live.
Canonized by The IT Crowd; annoying precisely because it works.
Have you tried turning it off and on again is a punchline about support laziness, and it is also genuinely the highest-yield first question in troubleshooting. That tension is worth taking seriously rather than resolving in either direction.
It works because a restart discards accumulated state, and accumulated state is where most faults live. Leaked memory, exhausted handles, a cached value that is now wrong, a thread stuck waiting on something that will never arrive, a driver in a condition its author never tested: none of these are in the code as written, they are in what the code became after running for a month. Restarting returns the system to the only configuration that was ever deliberately designed.
The professional objection is that it destroys the evidence, and that is correct and important. A restart fixes the symptom and guarantees the cause remains unknown, so the discipline is to capture state first, logs, memory, connection tables, whatever the situation allows, and then restart. Do it in that order and you get both the service back and the ability to prevent a recurrence. Do it in the other order and you will be here again, without having learned anything.