Amazon ECS Now Auto-Heals Broken Container Instances
Amazon ECS just got smarter at keeping your containers healthy—it now automatically detects when agent connectivity dies (thanks to EBS issues, network hiccups, or thermal events) and fixes it without you lifting a finger. For Fargate and ECS Managed Instances, it drains tasks and spins up replacements automatically; EC2 users get the health event to trigger their own recovery workflows. This magic works everywhere AWS operates, at zero extra cost.
source: [aws/whats-new]