For me, a long running and historically stable ECS service suddenly stopped.

Ordinarily this would result in failed health checks and auto service restart, but in this case, the service simply died without logs or alerts.