A readiness check periodically calls the /readyz endpoint to determine whether a Kubernetes pod is ready to receive traffic. Readiness requires the pod to be alive and operational enough to handle incoming requests. section 10.4.1
Readiness checks belong to the observability and pod-management layer of a deployed service. They complement liveness probes, which call /healthz to determine whether Kubernetes should restart a faulty process. section 10.4.1
BentoML service on Kubernetes
A BentoML Service includes built-in support for health checks and liveness/readiness probes, so a model-serving pod exposes the endpoints needed for Kubernetes monitoring without requiring custom endpoint implementation. section 10.4.1
Figure 10.7 BentoML’s built-in observability endpoints for K8s environments
Inspect the built-in observability endpoints used by BentoML Services in Kubernetes environments. section 10.4.1






