A liveness check determines whether an application can continue running. Kubernetes periodically calls the health endpoint used by the liveness probe; a failed check indicates an unrecoverable problem, such as exhausted memory or a deadlock, so Kubernetes restarts the pod. section 10.2
In Quarkus, SmallRye Health implements MicroProfile Health and exposes health information over HTTP. The aggregate endpoint is /q/health, while /q/health/live returns only liveness checks. Readiness is different: it reports whether the application can accept traffic, and a failure causes Kubernetes to withhold traffic rather than restart the pod. section 10.2 section 10.2.1
Use the probe types for different operational decisions:
- Liveness: restart the pod when the application is not expected to recover. section 10.2
- Readiness: stop sending traffic while a dependency or operating condition prevents request processing. section 10.2
- Startup: cover initialization until the application is ready, after which liveness and readiness take over. section 10.2
Expose the health endpoints
Add the SmallRye Health extension with `quarkus extension add smallrye-health`, run the application, and inspect `/q/health` or `/q/health/live`. Quarkus also provides a Dev UI Health UI for viewing checks during development. section 10.2.1
$ quarkus extension add smallrye-health
Install the Quarkus extension that supplies the SmallRye Health endpoints and health-check integration. section 10.2.1
