Question

A retail company runs a stateless web application across an auto scaling group behind a load balancer. During a recent incident, the load balancer continued sending traffic to instances that had crashed at the application layer even though the operating system was still running, resulting in intermittent HTTP 500 errors for customers. The architect wants to ensure that unhealthy instances are automatically detected and replaced without manual intervention. Which design change BEST addresses this requirement?

Accepted Answer

Configure application-layer (HTTP) health checks on the load balancer and set the auto scaling group to use the load balancer's health check to trigger instance replacement

Answer

Increase the minimum instance count in the auto scaling group so more capacity is always available

Answer

Enable TCP-based health checks on the load balancer to verify each instance is reachable on the listening port

Answer

Add a scheduled scaling policy that recycles all instances every 24 hours to clear crashed processes