Managing Kubernetes

Chapter 11. Monitoring Kubernetes

It’s all well and good to set up or use a Kubernetes cluster from a public cloud vendor. But without the right strategy for monitoring metrics and logs from that cluster and firing appropriate alerts when something goes wrong, the cluster that you have created is a disaster waiting to happen. Although Kubernetes makes it easy for developers to build and deploy their applications, it also creates applications that are dependent on Kubernetes for successful operation. This means that, when a cluster fails, your users’ applications often fail, as well. And, if a cluster fails too often, users lose trust in the system and begin to question the value of the cluster and its operators. This chapter discusses approaches to developing and deploying monitoring and alerting for your Kubernetes cluster to prevent this from happening. Additionally, we describe how you can add monitoring onto your cluster so that application developers can automatically take advantage of it for their own applications.

Goals for Monitoring

Before we step into the details of how to monitor your cluster, it’s important to go over the goals for this monitoring. All of the specifics of how to deploy and manage monitoring are in service of these goals, and thus a crystal clear sense of the why? will help in understanding the what.

Obviously, the first and foremost goal of monitoring is reliability. In this case, reliability is both that of the Kubernetes cluster and that of the ...