Versions & Update Configuration

This chapter describes how to configure the available Kubernetes/OpenShift versions and how to provide update paths for user clusters.

The list of selectable versions when specifying cluster name and Kubernetes version is defined in the spec.versions section in the KubermaticConfiguration CRD. This is also where updates are configured.

Default Versions

The list of default versions, is shown by the CRD example linked above, but it’s recommended to retrieve the actual list from our github repo itself.

They can be found in the docs directory.

Configuring Versions

The structure for Kubernetes and OpenShift versions is identical. Each contains

  • versions (array) is a list of user-selectable versions. These must be concrete semantic versions, wildcards or ranges are not supported.

  • default (string) is the default version for this cluster orchestrator, i.e. one of the items from versions.

  • updates (array) is a list of allowed update paths for a cluster. Each update consists of the following fields:

    • from (string) is a version (“v1.2.3”), a wildcard version (“v1.2.*") or version range (“v1.2.0-v1.3.0”) a cluster must match for this update to be allowed.
    • to (string) is a version, a wildcard version or version range that the cluster can be updated to.
    • automatic (bool) controls whether an update is performed immediately after applying the configuration. This is useful for force updates in case of security patch releases.
    • automaticNodeUpgrade (bool) controls whether worker nodes are updated as well. If this is left to its default (false), only the controlplane will be updated. When set to true, it implies automatic.

Each element of the two orchestrators can be overwritten independently, i.e. you can only override the list of allowed and default Kubernetes versions, while still relying on the default value for the update paths and all default settings for OpenShift by setting:

      versions: ['1.16.0', '1.16.2']
      default: '1.16.2'

It’s not possible to add or remove individual elements from the versions or updates arrays. You always have to specify the entire list in the KubermaticConfiguration.

Edit your KubermaticConfiguration either using kubectl edit or editing a local file and applying it with kubectl apply, the KKP Operator will reconcile the setup and after a few moments the new configuration will take effect.

Note that once you start overriding default values in your KubermaticConfiguration, you need to keep the settings up-to-date when upgrading KKP.