Monitoring on a cluster node

Monitoring is essential for Kubernetes administrators when it comes to getting a clear understanding of what’s going on in your Kubernetes cluster. You need to know all of the different metrics to help you get on track in terms of the health of your Kubernetes cluster components. You also need to make sure that your components are operating as expected and that all workloads that are deployed on your worker nodes are functional and have enough resources, such as CPU, memory, and storage. Moreover, you should also check whether any worker nodes are available and have sufficient resources to scale or schedule more workloads.

In Kubernetes, Metrics Server collects CPU/memory metrics and to some extent adjusts the resources needed by containers automatically. Metrics Server collects those metrics every 15 seconds from the kubelet agent and then exposes them in the API server of the Kubernetes master via the Metrics API. This process is described...