service health check

A regular automatic test, run by a load balancer, Auto Scaling group or container service, of whether a copy of a service can really handle requests.

In more detail

A load balancer sends a request to each target and stops routing to targets that fail; an Auto Scaling group or ECS service replaces copies that stay unhealthy. A good check tests the path users need, not just whether a process exists, because a running process can still be unable to serve. Repeated failures need logs and metrics to explain them, or automatic replacement just repeats a broken release.

Example

An Auto Scaling group with healthCheck set to load-balancer replaces instances the load balancer marks unhealthy.

Related terms

Taught in

References