Description

Ensure every Managed Kubernetes cluster runs a Kubernetes minor version that OVHcloud still supports, and that it is upgraded before its version reaches end of support.

Rationale

Unsupported Kubernetes versions stop receiving security patches, so known vulnerabilities in the control plane and nodes stay open. They may also block new features and integrations. Keeping clusters on a supported version keeps them patched and secure.

Impact

  • The cluster keeps receiving security fixes

  • Access to newer Kubernetes features

  • Upgrades roll through the nodes and must be tested for compatibility

Default Value

Clusters stay on the version they were created with until they are upgraded, although OVHcloud may force upgrades of versions at end of support.

Pre-Requisites

  • Permission to manage the Kubernetes cluster

  • Workloads tested against the next Kubernetes version

Test Plan

  1. Log in to the OVHcloud Control Panel at https://www.ovh.com/manager/.

  2. In the left-hand menu, select Public Cloud and choose your project.

  3. In the left-hand menu, select Managed Kubernetes Service, then click the cluster name to open it.

  4. Check the Kubernetes version shown on the cluster's Service tab.

  5. Compare it with the versions supported by OVHcloud.

  6. If the version is unsupported or close to end of support, follow the implementation steps.

Implementation Steps

  1. Log in to the OVHcloud Control Panel at https://www.ovh.com/manager/.

  2. In the left-hand menu, select Public Cloud and choose your project.

  3. In the left-hand menu, select Managed Kubernetes Service, then click the cluster name to open it.

  4. In the Management section, click Upgrade to the next minor version and confirm (allow about 5 minutes per node).

  5. Wait until the Control Panel no longer shows that the cluster is being upgraded.

  6. Repeat until the cluster is on a supported version, then test your workloads.

Backout Plan

  1. Once an upgrade is performed, you cannot revert to the previous version. If needed, create a new cluster on the previous version and redeploy the workloads.

Reference