Problem

• importing the local-cluster and creating the addon is extra steps that slows adoption.

Solution

• We should create the local-cluster import resource and the hypershift Addon if it doesn't exist, when the hypershift flag is enabled.
• This should work with MCE or ACM
  • This should not remove the local cluster, that will be manual cleanup and makes sure we don't cause problems for ACM
  • This means we can't just include the two resources in backplane-pipeline chart, the addon itself should create the resources if they are missing.

Resources to create

local-cluster:
apiVersion: cluster.open-cluster-management.io/v1
kind: ManagedCluster
metadata:
  labels:
    local-cluster: "true"
  name: local-cluster
spec:
  hubAcceptsClient: true
  leaseDurationSeconds: 60
 
addon:
apiVersion: addon.open-cluster-management.io/v1alpha1
kind: ManagedClusterAddOn
metadata:
  name: hypershift-addon
  namespace: local-cluster
spec:
  installNamespace: open-cluster-management-agent-addon
 
The one catch is that the Hypershift controller will not be able to work with AWS until the S3 secret is created. Agent, Azure and Kubevirt will be good to go though.

In the UX flow we will want to have a warning about this.

There is a script that does this step today: https://github.com/stolostron/hypershift-deployment-controller/blob/main/samples/quickstart/start.sh