Book Image

kubectl: Command-Line Kubernetes in a Nutshell

By : Rimantas Mocevicius
Book Image

kubectl: Command-Line Kubernetes in a Nutshell

By: Rimantas Mocevicius

Overview of this book

The kubectl command line tool lets you control Kubernetes clusters to manage nodes in the cluster and perform all types of Kubernetes operations. This introductory guide will get you up to speed with kubectl in no time. The book is divided into four parts, touching base on the installation and providing a general overview of kubectl in the first part. The second part introduces you to managing Kubernetes clusters and working with nodes. In the third part, you’ll be taken through the different ways in which you can manage Kubernetes applications, covering how to create, update, delete, view, and debug applications. The last part of the book focuses on various Kubernetes plugins and commands. You’ll get to grips with using Kustomize and discover Helm, a Kubernetes package manager. In addition to this, you’ll explore how you can use equivalent Docker commands in kubectl. By the end of this book, you’ll have learned how to install and update an application on Kubernetes, view its logs, and inspect clusters effectively.
Table of Contents (16 chapters)
Section 1: Getting Started with kubectl
Section 2: Kubernetes Cluster and Node Management
Section 3: Application Management
Section 4: Extending kubectl

Cordoning nodes

Let's suppose we are going to run an app's load test and we want to keep a node away from the load test. In the node list that we saw in the Getting a list of nodes section, we have three nodes, and they are all in the Ready state. Let's pick one node, gke-kubectl-lab-default-pool-b3c7050d-8jhj, which we do not want new pods to be scheduled on.

kubectl has a command called cordon, which allows us to make a node unschedulable:

$ kubectl cordon -h
Mark node as unschedulable.
  # Mark node "foo" as unschedulable.
  kubectl cordon foo
      --dry-run='none': Must be "none", "server", or "client". If client strategy, only print the object that would be
sent, without sending it. If server strategy, submit server-side request without persisting the resource.
  -l, --selector='': Selector (label query) to filter on