Kuryr and Kubernetes Probes - pt 2

Adding Kubernetes probes to Kuryr-Kubernetes

Over the last weeks since my last post I continued working on improving the management that Kubernetes does of Kuryr-Kubernetes pods.

Kuryr Controller pod

For the Controller pod, besides having added the readiness probe, I started adding a liveness probe, which will assure that the Controller is alive and healthy to perform its operations.

The liveness probe will work by sending an HTTP request to the Health Manager Server who will check the health state of Kuryr-Controller components that were registered to be checked. If all the components are healthy it will return success status and will be considered alive, otherwise a failure will be registered and the container will restart.

Kuryr CNI pod

I also started creating another similar Health Manager Server that will be running inside CNI container alongside the CNI daemon and will be monitoring it. For now, the manager is only checking Kuberntes healthy and so providing readiness to the probe. Later, I will check the status of CNI components.

A patch is on review for the controller liveness and another one was opened in WIP for the CNI health checks.