Health Check
Health check endpoints expose a service's liveness and readiness, letting orchestration platforms restart or deregister unhealthy instances.
Purpose
Let the orchestration platform automatically detect and recover failed instances, improving availability.
Mechanism
Liveness: whether the service is alive (restart on failure). Readiness: whether it is ready to serve traffic (deregister from the load balancer on failure). Startup: don't kill during startup. K8s probes support HTTP/TCP/Exec checks.
Use cases
- Failure detection
- Automatic recovery
- Load balancing