Upgrading Self-Managed Nodes to a Newer Kubernetes Version by Replacing an Existing Self-Managed Node

Find out how to upgrade the version of Kubernetes running on a self-managed node in an enhanced cluster created with Container Engine for Kubernetes.

You can 'upgrade' the version of Kubernetes running on a self-managed node by replacing the original self-managed node with a new self-managed node running the newer Kubernetes version. Having drained the original self-managed node to prevent new pods starting and to delete existing pods, you can then terminate the compute instance hosting the original self-managed node.

Note that when creating the new self-managed node, it is your responsibility to specify an image containing a Kubernetes version that complies with the Kubernetes version skew support policy described in the Kubernetes documentation. Container Engine for Kubernetes does not check that the Kubernetes version in the image you specify for the new self-managed node is compatible with the Kubernetes version running on the cluster's control plane nodes. See Cluster Requirements.

To 'upgrade' the version of Kubernetes running on a self-managed node by replacing the original self-managed node with a new self-managed node:

1. Create a new compute instance to host the new self-managed node:
   1. Open the navigation menu and click Compute. Under Compute, click Instances.
   2. Follow the instructions in the Compute service documentation to create a new compute instance. Note that appropriate policies must exist to allow the new compute instance to join the enhanced cluster. See Creating a Dynamic Group and a Policy for Self-Managed Nodes.
   3. In the Image and Shape section, click Change image.
   4. Click My images, select the Image OCID option, and then enter the OCID of the OKE Oracle Linux 7 (OL7) or Oracle Linux 8 (OL8) image running the newer Kubernetes version. See Image Requirements.
   5. Click Show advanced options, and on the Management tab, select the Paste cloud-init script option.
   6. Copy and paste the cloud-init script containing the Kubernetes API private endpoint and base64-encoded CA certificate into the Cloud-init script field. See Creating Cloud-init Scripts for Self-managed Nodes.
   7. Click Create to create the compute instance to host the self-managed node.

   When the compute instance is created, it is added as a self-managed node to the cluster with the Kubernetes API endpoint that you specified.

2. Verify that the self-managed node has been added to the Kubernetes cluster and confirm the node's readiness status by entering:

   kubectl get nodes

   For example:

   kubectl get nodes
   
   NAME           STATUS   ROLES    AGE   VERSION
   10.0.103.170   Ready    <none>   40m   v1.25.4

3. Confirm that labels have been added to the node and set as expected by entering:

   kubectl get node <node-name> -o json | jq '.metadata.labels'

   For example

   kubectl get node 10.0.103.170 -o json | jq '.metadata.labels'
   
   {
   ...
   "displayName": "oke-self-managed-node",
   "oci.oraclecloud.com/node.info.byon": "true",
   ...
   }

4. If there are labels attached to the original self-managed node and those labels are used by selectors (for example, to determine the nodes on which to run pods), then use the kubectl label nodes command to attach the same labels to the new self-managed node. See Assigning Pods to Nodes in the Kubernetes documentation.

5. Prevent new pods from starting, and delete existing pods, on the original self-managed node by entering:

   kubectl drain <node_name>

   For more information:

   • about using kubectl, see Accessing a Cluster Using Kubectl
   • about the drain command, see drain in the Kubernetes documentation

   Recommended: Leverage pod disruption budgets as appropriate for your application to ensure that there's a sufficient number of replica pods running throughout the drain operation.

6. When you have drained the original self-managed node and pods are running on the new self-managed node, terminate the compute instance hosting the original self-managed node. See Terminating an Instance.