Health Check Endpoints

Health check endpoints are dedicated HTTP routes that report whether an application instance is functioning correctly. They serve as the primary mechanism for load balancers, container orchestrators, and monitoring systems to determine whether to route traffic to an instance, restart it, or trigger alerts. Well-designed health checks are critical for automated recovery and zero-downtime deployments.

Kubernetes distinguishes between three types of probes: liveness (is the process alive and not deadlocked -- restart if failing), readiness (can this instance serve traffic -- remove from load balancer if failing), and startup (has initial loading completed -- don't check liveness until passing). Each serves a different purpose and should check different things. A liveness check should be lightweight (is the event loop responsive), while a readiness check should verify connectivity to critical dependencies (database, cache, external APIs).

Health check responses should include structured JSON with overall status, individual dependency statuses, and relevant metadata (version, uptime, response times). Checks should have timeouts shorter than the probe interval to prevent cascading failures. Avoid checking non-critical dependencies in readiness probes to prevent unnecessary cascading unavailability. Consider providing a detailed /health endpoint for monitoring dashboards and a lightweight /healthz for high-frequency load balancer polling.