Kubernetes Version Upgrade

You can upgrade the version of a Kubernetes cluster running on Linux operating system. You can perform version upgrade operations from Kubernetes 1.18 to 1.30, transition from Docker to Containerd, and manage necessary files.

When upgrading Kubernetes version, each minor version between the current and target versions (e.g., from 1.25 to 1.26, then from 1.26 to 1.27, etc.) must be upgraded sequentially.

Kubernetes: v1.18 → Current version, v1.30 → Target version

Required Files

Kubernetes Files

v1.18 - v1.30 → Kubectl, kubeadm, kubelet files

Download

Kubernetes CNI

Installing once is sufficient when transitioning from Kubernetes 1.18 to 1.19. Installation is not necessary in versions 1.19 and above.

Containerd

When transitioning to Kubernetes 1.24 version, transition to containerd usage should be made. Containerd should be 1.6+ and docker should be removed.

Download

Version Upgrade

The version upgrade process consists of “1.18-1.19”, “1.20-1.23”, and “1.24-1.30” sections. In each section, the version upgrade process and solutions to possible errors are addressed.

1.1) Upgrade from 1.18 to 1.19

Very Important

A flannel issue may occur when transitioning from 1.18 to 1.19. Check node status with the following command.

kubectl get nodes

If node status is NotReady, check logs on the relevant server with the following command.

journalctl -xeu kubelet

Solution: Transfer the flannel file from any (master or worker) server to the server you are upgrading using scp command.

scp /opt/cni/bin/flannel <USERNAME>@<NODE_IP>:/opt/cni/bin

When you check the node status, you should see “READY”.

Upgrade for Control Plane (Master) Node

Switch to the server that is Master(control-plane) and apply the following steps.

1. Check current versions

   Check kubeadm, kubelet, and kubectl versions before starting the upgrade process:

   kubeadm version
   kubelet --version
   kubectl version --client
   
2. Upgrade Kubeadm version

   Remove package hold for Kubeadm:

   sudo apt-mark unhold kubeadm
   

   Go to the folder where you extracted the file you downloaded from fss.apinizer.com:

   cd 1.19.3
   

   Install the .deb extension kubeadm file found inside:

   sudo dpkg -i kubeadm_1.19.3-00_amd64.deb
   sudo apt-mark hold kubeadm
   
3. Put node in maintenance mode

   The drain command below puts the specified Kubernetes node in maintenance mode and transfers pods to other nodes. If you do not have an alternative master node for your Master (control-plane) node, this command will cause system interruption. You may not apply this operation for clusters with a single node.

   kubectl drain <MASTER_NODE_NAME> --ignore-daemonsets
   
4. Perform Kubeadm upgrade

   Upgrade Kubeadm to the installed v1.19.3 version with upgrade command:

   # For Master (control-plane):
   # change the version part for each version
   sudo kubeadm upgrade apply v1.19.3
   

   You should see a message indicating that the operation was successful as output on the screen.

5. Make node available again

   kubectl uncordon <MASTER_NODE_NAME>
   
6. Upgrade Kubelet and Kubectl versions

   sudo apt-mark unhold kubelet kubectl
   sudo dpkg -i kubelet_1.19.3-00_amd64.deb
   sudo dpkg -i kubectl_1.19.3-00_amd64.deb
   sudo apt-mark hold kubelet kubectl
   
7. Restart Kubelet service

   systemctl daemon-reload
   systemctl restart kubelet
   
8. Perform version check

   To view the version of the Master (control-plane) node:

   kubectl get nodes
   

   Check that the upgrade operation was successful as the version.

Upgrade for Worker Node

Worker node version upgrade is also performed with similar steps.

1. Put worker node in maintenance mode

   The drain command below puts the specified Kubernetes node in maintenance mode and transfers pods to other nodes. If you do not have an alternative node for your Worker node, this command will cause system interruption. You may not apply this operation for clusters with a single worker node.

   The command is run on the master(control-plane) server and initiates maintenance and pod evacuation for the worker node.

   kubectl drain <WORKER_NODE_NAME> --ignore-daemonsets
   
2. Switch to worker node server

   Switch to the server where the worker node is located and perform kubeadm upgrade operation.

   Go to the folder where you extracted the file you downloaded from fss.apinizer.com:

   sudo apt-mark unhold kubeadm
   sudo dpkg -i kubeadm_1.19.3-00_amd64.deb
   sudo apt-mark hold kubeadm
   
3. Perform upgrade for worker node

   sudo kubeadm upgrade node
   
4. Upgrade Kubelet and Kubectl versions

   sudo apt-mark unhold kubelet kubectl
   sudo dpkg -i kubelet_1.19.3-00_amd64.deb
   sudo dpkg -i kubectl_1.19.3-00_amd64.deb
   sudo apt-mark hold kubelet kubectl
   
5. Restart Kubelet service

   systemctl daemon-reload
   systemctl restart kubelet
   
6. Make node available again

   Run the following command on the master(control-plane) node to make the node available again:

   kubectl uncordon <WORKER_NODE_NAME>
   
7. Perform version check

   To view the version of the Worker node:

   kubectl get nodes
   

1.2) Upgrade from 1.20 to 1.23

Versions 1.20-1.23 are updated one by one. The steps above are repeated for each version.

1.3) Upgrade from 1.24 to 1.30

Very Important

When transitioning to Kubernetes 1.24 version, transition to containerd usage should be made instead of Docker. Containerd should be 1.6+ and docker should be removed.

Versions 1.24-1.30 are updated one by one. The steps above are repeated for each version. However, containerd installation must be done when transitioning to 1.24.

Containerd Installation (for Kubernetes 1.24)

When transitioning to Kubernetes 1.24 version, transition to containerd usage should be made instead of Docker.

Containerd Installation

You can download the necessary file for Containerd installation.

See the relevant documentation pages for Containerd installation and Docker transition operations.