03-install-kubernetes-masters-and-nodes

Installation, Configuration & Validation: Install Kubernetes Masters and Nodes

Back to Certified Kubernetes Administrator (CKA) Tutorial

note: Includes the following additional topics under Installation, Configuration & Validation.

• Design a Kubernetes Cluster

• Know where to get the Kubernetes release binaries

• Provision underlying infrastructure to deploy a Kubernetes cluster

• Choose a network solution

• Choose your Kubernetes infrastructure configuration

• Install and use kubeadm to install, configure, and manage Kubernetes clusters

Test Preparation: Relevant Kubernetes Documentation

In our example, we added a Worker Node as part of the intitial install. In the exam, we need to be prepared to add a Worker Node after-the-fact using the instructions at Creating a single control-plane cluster with kubeadm; one thing document fails to mention is that the kube-apiserver runs on port 6443. Search for install kubeadm.

It turns out that kubeadm has an option to printout the join command when creating a new token (use help, but believe option is --print-join-command).

While most of the K8s PKI certificates are in the folder /etc/kubernetes/pki, kublet keeps them in a separate (hard to find) location; defaults to /var/lib/kubelet/pki.

Another important file location is /etc/cni/net.d as it contains the CNI Addon configuration file.

Script

There are compelling reasons to use Amazon's EKS instead of trying to install your own cluster from scratch; mostly EKS has deep integrations with other AWS services out of the box and is HA. But we are here to learn...

Also, while one can install learning environments on one's workstation, e.g,. Minikube or microK8s, we want to install in a more "real" environment.

There are a number of tools that automate the creation of clusters (much like EKS), e.g.:

• kops: The easiest way to get a production grade Kubernetes cluster up and running

• KRIB: Kubernetes Rebar Integrated Bootstrap content pack

• Kubespray: Deploy a Production Ready Kubernetes Cluster

As we are here to learn, we are going to manually install a Kubernetes cluster using the kubeadm tool.

Installation

For simplicy, we will use a network configuration with a single public subnet, e.g., the default AWS VPC and subnets.

We start with using AWS console to launch two EC2 instance.

• t3.medium

• Ubuntu LTS

• Security Group: TCP/22 from anywhere, All traffic from the Security Group itself

Installation starts with installing kubeadm on both instances. Steps generally involve:

• Bridge: kernel and config

• Container Runtime: Docker

• kubeadm, kublet, and kubectl

To keep it simple, we will follow the instructions in Creating a single control-plane cluster with kubeadm. This configuration has one instance as a Control Plane Node (aka, Master Node) and one as a Worker Node.

The actual installation, planning for the Calico Pod Network, is the following command followed with the generated instructions:

kubeadm init --pod-network-cidr 192.168.0.0/16

As we will learn, many Kubernetes objects, e.g., Pods, are associated with an IP address and these addresses are drawn from this CIDR network. Notice this address range is completly separate from the IP addresses in the VPC (important as they are independently managed).

The next step is to install Calico Pod Network Addon.

Finally, on second host, follow instructions to join as Worker Node.

Validation On Each Node

note: All the following quotes and diagram are from Kubernetes Components.

Container Runtime (runs as an OS service, e.g., Docker)

The container runtime is the software that is responsible for running containers.

Confirm operational:

systemctl status docker

Kubelet (runs as an OS service)

An agent that runs on each node in the cluster. It makes sure that containers are running in a Pod.

Confirm operational:

systemctl status kubelet

kube-proxy (runs as a Pod)

maintains network rules on nodes. These network rules allow network communication to your Pods from network sessions inside or outside of your cluster

Confirm operational (run from Master):

kubectl get pods --all-namespaces --field-selector spec.nodeName=ip-172-31-42-199

note: Your Node name will be different.

Validation On the Master (or Control Plane) Node

kube-apiserver (runs as a Pod)

The API server is a component of the Kubernetes control plane that exposes the Kubernetes API. The API server is the front end for the Kubernetes control plane.

Confirm operational (for following Pods too):

kubectl get pods --all-namespaces --field-selector spec.nodeName=ip-172-31-47-139

note: Your Node name will be different.

note: Observe the CoreDNS and Calico Addons.

kube-scheduler (runs as a Pod)

Control plane component that watches for newly created Pods with no assigned node, and selects a node for them to run on.

kube-controller-manager (runs as a Pod)

Control Plane component that runs controller processes.

cloud-controller-manager (runs as a Pod)

A Kubernetes control plane component that embeds cloud-specific control logic.

etcd (runs as a Pod)

Consistent and highly-available key value store used as Kubernetes’ backing store for all cluster data.