Community

Blog Events Webinars Tutorials Forum
  1. Blog
  2. Events
  3. Webinars
  4. Tutorials
  5. Forum
Create Account
×
Community Blog Use Kubeadm to Create a Kubernetes Cluster

Use Kubeadm to Create a Kubernetes Cluster

In this tutorial, you will learn how to install a Kubernetes cluster on Alibaba Cloud installed with Kubeadm, which is a tool that make managing Kubernetes even easier.

By Alex Mungai Muchiri, Alibaba Cloud Community Blog author

Kubernetes does not have to be complicated and messy. You can achieve quick turnaround and configure a Kubernetes cluster easily using the Kubeadm toolkit, which is a tool to automate the installation and configuration of Kubernetes components. The tool can help set up an executable cluster with minimal working components in place. It handles installations such as the API server, Controller Manager, and Kube DNS.

In this tutorial, you will learn how to install a Kubernetes cluster on an Alibaba Cloud ECS instance installed with Ubuntu 16.04 with Kubeadm. The purpose of this tutorial is just to give you a simple test environment.

Prerequisites

Before you begin this tutorial, make sure that you have the following items:

  • Three Alibaba Cloud ECS instances installed with Ubuntu 16.04 with a minimum of 1GB RAM.
  • Root access to the three servers
  • Experience launching containers from Docker images.

Specifically, the three servers or hosts should have familiar names, such as kube-1, kube-2 and kube-3 as shown below:

Server Hostname
1 kube-01
2 kube-02
3 kube-03

In this tutorial, we will create roles for the three hosts as a master and two workers. While not essential to the configuration of Kubernetes clusters, keep in mind the following:

Assigned roles Hostname
Master kube-01
Worker kube-02
Worker kube-03

Now proceed with the server preparation.

Procedure

Follow the steps outlined below to install a Kubernetes clusters on Alibaba with Kudeadm.

Prepare the Server

In this step, we'll prepare our servers for the configuration of our Kubernetes cluster. The first step in preparing the server is ensuring your virtual servers have unique hostnames. For this tutorial, I recommend that you use kube-1, kube-2, and kube-3. You'll also need MAC addresses and some other unique identifiers. In reality, Alibaba Cloud automatically assigns unique identifiers to your servers so to ensure the hostnames are also unique.

The next step in preparing the server is enabling the ports 6443 and 10250 for the Kubernetes API Server and Kubelet API respectively. As a requirement, the commands below should be run with the root user.

Run the command:

firewall-cmd --get-active-zones

Next, activate the ports by running the commands below:

firewall-cmd --zone=public --add-port=6443/tcp --permanent
firewall-cmd --zone=public --add-port=10250/tcp --permanent

Finally, you will need to switch off SWAP partitioning by running the command below:

sudo swapon -s
sudo swapoff -a

The first command checks the swap list while the next one disables it. Repeat the commands for all the three server instances and the proceed to the next step. However, for permanent disabling, you need to access the /etc/fstab file in your favorite editor like so:

sudo nano /etc/fstab

Next, you'll want to find the line below and place a # comment like so:

#/dev/mapper/hakase--labs--vg-swap_1 none     swap    sw        0       0

Save, exit and reboot system with the below command:

sudo reboot

Install Docker

We'll step up Docker for container runtime and iinstall Docker to run containers for future projects and of course, for purposes of demonstrating the lessons of this tutorial. Installing Docker in an Ubuntu server is as easy as running the commands below in all three servers:

curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add ¨C
sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable"
sudo apt-get update
apt-get install -y docker-ce

It is important to note that Kubernetes supports the latest Docker Engine version and there are potential conflicts if a newer version is installed. So, ensure that you check the supported version for the installed Kubernetes release.

The command below checks the Docker version:

docker version

When you have the right version installed, you may need to hold it so that it is not updated by mistake and hence become incompatible with Kubernetes.

sudo apt-mark hold docker-ce

Enable and start Docker using the commands below:

sudo systemctl enable docker
sudo systemctl start docker

Now we have our container runtime installed and ready to deploy the Kubeadm toolkit

Install Kubeadm, Kubelet and Kubectl

In this step, we'll need to SSH into each of the three servers that make up our cluster and install the three components. Again, you will need to login as root to be able to run these commands.

Accordingly, run the commands below in Kube-1, Kube-2 and Kube-3:

apt-get update && apt-get install -y apt-transport-https curl
curl -s https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add -
cd /etc/apt
sudo nano sources.list.d/kubernetes.list

Paste the line below in the file:

deb http://apt.kubernetes.io/ kubernetes-xenial main

Then, run the commands below:

apt-get update
apt-get install -y kubelet kubeadm kubectl

However, we should ensure that Kubernetes does not get updated to a version that would be incompatible with Docker. We shall update the package manager to exclude Kubernetes from potential updates by running the command below:

sudo apt-mark hold kubelet kubeadm kubectl

Now that we have ensured that the packages remain in sync, we shall proceed to configure the cgroup driver.

Configure the default cgroup Driver in kube-1

This is not a mandatory step and whether to proceed with this step or to bypass it will depend on the results of the command below. In kube-1 server, run the command:

sudo docker info | grep -i cgroup

If the output of this command is something of this sorts, then you will need cgroup configuration.

Cgroup Driver: cgroupfs

Now we'll modify the default kubelet default file and include the Cgroup driver that we have seen above. In your favourite editor (I use nano), run the command below:

sudo nano /etc/default/kubelet

Include the following line in the file:

KUBELET_KUBEADM_EXTRA_ARGS= - cgroup-driver=< cgroupfs >

Now you should restart Kubelet by running the commands below:

systemctl daemon-reload
systemctl restart kubelet

Great, we are all set for the master configuration in the next step.

Initialize Master Node

Now SSH into the kube-1 server to initiate the master node. The command below will be all we need:

kubeadm init --pod-network-cidr=192.168.100.10/6433

During the server preparation, we enabled the ports 6443 and 10250. Now, be sure to replace the 10.244.0.0/16 in the command above with your IP/port combination.

Let us breakdown the kubeadm init command above:

  • --pod-network-cidr: This will define the Pod network segment for your cluster.
  • --apiserver-advertise-address: This argument defines the IP address to be used by the Kubernetes API and could of any type.
  • --apiserver-cert-extra-sans: This argument serves to define the IP or domain to be included in the certificate sans, to include more information apart from the host machine hostname and IP.

When executed, you should see numbers running down the black screen and then finally a 'Your Kubernetes master has initialized successfully!' message. You should also get a line that you will run in the worker nodes to link them to the master.

Now, follow the steps below to make modifications to the user shell environment if you are using a non-root user to run these commands:

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

As was mentioned earlier, you should also get a line that you will run in the worker nodes to link them to the master.

kubeadm join 192.168.100.10/6433--token bkz1q4.yuevvhvve90jk --discovery-token-ca-cert-hash sha256: F3409C6C295F87249D5C08DA11791F3452950A0ACE646DF2BAD06514940DF847

This line is very useful and without it, your worker nodes cannot be enjoined to the cluster. Copy it in preparation for linking your two workers. However, if it is more than24 hours since it was generated, you will need to regenerate using the command below:

kubeadm token create --print-join-command

Great, let now deploy the cluster networking in the next step

Deploy Kubernetes Cluster Networking

Cluster networking facilitates communication between pods, containers and between external services and nodes. The chosen model is the Flannel, which we will apply by running the command below:

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

So far we have set up the cluster networking and you can verify that by running the command below:

kubectl get deployments --all-namespaces

You should be able to see your master listed. Now we shall add the workers:

Add Worker

As we initiated the master node, we received an initializing command. In the remaining servers (kube-1 and kube-2), we will run the command below to add them as workers:

kubeadm join 192.168.100.10/6433--token bkz1q4.yuevvhvve90jk --discovery-token-ca-cert-hash sha256: F3409C6C295F87249D5C08DA11791F3452950A0ACE646DF2BAD06514940DF847

As a reminder, you will need to have installed Docker, kubeadm and kubernetes in the worker nodes prior to this step. Once the command runs, you should see a This node has joined the cluster message.

Repeat the command to all hosts.

Verify the Cluster

Clusters could fail at the setup for any number of reasons. It is thus important to verify that all nodes are running properly. Check from the master node by this SSH command:

ssh ubuntu@master_ip

Next, run the below command to retrieve the cluster's status:

kubectl get nodes

You should receive a similar output:

Output
NAME      STATUS    ROLES     AGE       VERSION
Kube-1    Ready     master     1d        v1.10.1
Kube-2    Ready     <worker>   1d        v1.10.1 
Kube-3    Ready     <worker>   1d        v1.10.1

If the output indicates that all nodes are ready, we can proceed to run some workloads. If they are not, wait for five minutes and try again. If that fails, repeat the installation until you succeed.

Conclusion

This tutorial has guided you on how to set up a Kubernetes cluster in Ubuntu with Kubeadm. You can try deploying new services to the cluster to get comfortable working with Kubernetes. Kubernetes offers a great advantage with its functionalities and features, which you can learn from the Kubernetes Official Documentation. Alibaba Cloud provides a stable and reliable cloud platform to run these containerized applications made by Kubernetes.

Do you have an Alibaba Cloud account? Sign up for an account and try over 40 products for free worth up to $1200. Get Started with Alibaba Cloud to learn more.

Docker Container Tech Share Serverless Kubernetes Tutorials Kubeadm
0 0 0
Share on

Read previous post:

Deploy IoT Applications with Node-RED on Alibaba

Read next post:

Setting up and Troubleshooting Your Nginx Server on Alibaba Cloud

Alex

53 posts | 8 followers

You may also like

Alex

53 posts | 8 followers

Related Products

More Posts by Alex

See All