[Operational Awareness] Review the Kubernetes readiness probe and liveness probe of KIM

[tobiscr]

Description

We have ensure the KIM pod exposes a Readiness probe (system ready for taking over load) and a Liveness probe (system unhealthy and needs restart).

The Kubebuilder framework provides already pre-implemented probes (see https://book.kubebuilder.io/cronjob-tutorial/empty-main). But we have to ensure their implementation is suitable for KIM's context and giving a proper indication about KIMs internal readiness/health state.

AC:

• KIM supports a Readiness probe which gives a clear indication to Kubernetes when the pod is ready to process requests. Requirements to be ready:

  • Configuration validation successfully finished
  • Readiness probe should indicate ok when bootstrap is completed

• KIM supports a Liveness probe which indicates the internal health state of KIM to Kubernetes (and could trigger a restart of the pod whenever needed). Requirements for healthiness:

  • Liveness probe should always indicating ok, KIM has no special case which needs to be monitored and should trigger a restart of the pod (see screenshot of current implementation)

    

Reasons

Support Kubernetes mechanism for indicating the health and readiness state of a pod.

Attachments