Backend & systems·Resilience and failure handling
you deliberately kill an instance in production during working hours to find out whether the failover actually works.
Chaos engineering
Also calledfault injection, game day
Injecting controlled failures into a running system to verify that its resilience mechanisms work before an incident tests them. The value is in disproving assumptions: the retry that was never exercised, the replica that was never promoted. Doing it without a hypothesis, a blast radius limit and a stop button is not chaos engineering, it is an outage.