In this tutorial we're going to configure a multi-node cluster on Azure cloud from the ground up (without using the Azure's pre-configured Kubernetes platform). We'll also install all the required software on each VM such as Docker, Kubelet, Kubeadm and Kubectl. The cluster will consist of 1 master and 1 worker node as pointed out below running on CentOS 7 VMs.

Master: 2 vCPUs - 4GB Ram
Worker: 2 vCPUs - 4GB Ram

In addition to that, we'll also be using Flannel for managing the network. However you may use similar tools of your choice such as Weave Net for this purpose.

Step 1:Creating the VMs for master and worker nodes in Azure cloud dashboard.

Click on Virtual machines and select Create and then select Azure Virtual Machine in Virtual Machines tab to begin creating the master node.

In the Basics tab do the following:

Select the default Resource Group or create a custom one. In this example, the default NetworkWatcherRG will be selected.
Provide a name for the VM.
Select a region
Select CentOS 7.9 for the Image.
Select 2 CPUs and 4GB ram for Size of the RAM
SSH Access:
Add a username for SSH
Use existing public key and add an SSH key.
Select inbound ports option on port 22.

Keep rest of the options to the default (on other tabs) and click on [Review + create] option and then hit Create to deploy the VM.

Repeat the same steps above to create the worker node and once this is done, we should have two VMs appearing on the dashboard along with the Resource Group as shown below.

Step 2: Accessing the VMs using SSH and switch to superuser

Grab public IPs from the VMs created above by clicking into each VM.

Then SSH into the VMs using usernames and IPs as below for example (make sure to say yes for the fingerprint).

ssh yrol@20.37.251.35

Now access the root user

sudo su

Step 3: Disable - Firewall | Swap | SELinux

Note: Execute on all nodes (master & worker)

Disable Firewall - please note this is not the recommended way of doing in a production ready environment.

# Disable Firewall
systemctl stop firewalld
systemctl disable firewalld

Alternatively, ensure the ports [6443 and 10250] are open instead of disabling the firewall as above.

Check firewall status to make sure its disabled

# Check firewall status
firewall-cmd --state

Disable Swap - this needs to be done due to the performance issues discussed in here.

# Disable Swap
swapoff -a
sed -i.bak -r 's/(.+ swap .+)/#\1/' /etc/fstab

Disable SELinux - performing this due to the incompatibility issues between Kubelet and SELinux.

# Disable SELinux
setenforce 0
sed -i 's/enforcing/disabled/g' /etc/selinux/config

Step 4: Download & Install - Docker | Kubelet | Kubeadm | Kubectl

Note: Execute on all nodes (master & worker)

Adding a Kubernetes Repository

# Kubernetes Repository
cat <<EOF > /etc/yum.repos.d/kubernetes.repo
[kubernetes]
name=Kubernetes
baseurl=https://packages.cloud.google.com/yum/repos/kubernetes-el7-x86_64
enabled=1
gpgkey=https://packages.cloud.google.com/yum/doc/yum-key.gpg https://packages.cloud.google.com/yum/doc/rpm-package-key.gpg
exclude=kube*
EOF

Installing Docker | Kubelet | Kubeadm | Kubectl

yum update -y
yum install -y docker kubeadm kubelet kubectl --disableexcludes=kubernetes

Start and enable docker and kubectl

systemctl enable docker && systemctl start docker
systemctl enable kubelet && systemctl start kubelet

Enable IP table filtering to make sure the proxying is working correctly.

# For CentOS and RHEL
cat <<EOF >  /etc/sysctl.d/k8s.conf
net.bridge.bridge-nf-call-ip6tables = 1
net.bridge.bridge-nf-call-iptables = 1
EOF

Enable IP forwarding.

sysctl net.bridge.bridge-nf-call-iptables=1
sysctl net.ipv4.ip_forward=1
sysctl --system
echo "1" > /proc/sys/net/ipv4/ip_forward

Reload the daemon and restart the Kubelet

systemctl daemon-reload
systemctl restart kubelet

Step 5: Configure Kubernetes master node

Initializing the master node

kubeadm init --pod-network-cidr=10.240.0.0/16

After this step is successfully completed, make sure to note down the join command outputted in the console which consist of the token (valid for only 24 hours). We'll be using this command in our worker node to join with the master node. For instance, I had the following join command as an output in the console.

kubeadm join 10.0.0.5:6443 --token 8tfpf2.pj4zsj6vi5x0jj6j --discovery-token-ca-cert-hash sha256:d8b7c6f7f8649920cdfdcc906dd80e3b6a6c093b19e7bc844c51d0b64ae7ca8e

If you want to run kubectl as "regular" user. Then, execute below.

mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

Installing Flannel network plug-in for cluster network

kubectl apply -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml

Step 5: Join worker nodes to the cluster

Note: make sure to get the exact join command from previous kubeadm init command output (with the token). Ex:

kubeadm join 10.0.0.5:6443 --token 8tfpf2.pj4zsj6vi5x0jj6j --discovery-token-ca-cert-hash sha256:d8b7c6f7f8649920cdfdcc906dd80e3b6a6c093b19e7bc844c51d0b64ae7ca8e

Once the above command is executed successfully on the worker node, we should see an output like below.

Step 6: Testing

Note: Execute on master node

Displaying the nodes - execute on master node

kubectl get nodes

As shown, we've the master and worker nodes configured above are in Ready status.

Deploying a sample Nginx app (consist of 3 replicas)

kubectl apply -f https://raw.githubusercontent.com/kubernetes/website/master/content/en/examples/controllers/nginx-deployment.yaml

Displaying the PODs

kubectl get po -o wide

As shown, the 3 PODs belonging to Nginx app we just deployed is being run on the worker-1 node.