Deployment cannot be synced to cluster about kcp

[y-taka-23]

I ran into the same phenomenon.

Surely it is because the current implementation of the deployment-splitter controller detects the diff in and update the Status only. (pkg/reconciler/deployment/controller.go)

if !equality.Semantic.DeepEqual(previous.Status, current.Status) {
	_, uerr := c.client.Deployments(current.Namespace).UpdateStatus(ctx, current, metav1.UpdateOptions{})
	return uerr
}

On the other hand, in the case of a single downstream cluster, its reconciliation loop only adds a label cluater: your-cluster-name (pkg/reconciler/deployment/deployment.go)

if len(cls) == 1 {
	// nothing to split, just label Deployment for the only cluster.
	if root.Labels == nil {
		root.Labels = map[string]string{}
	}

	// TODO: munge cluster name
	root.Labels[clusterLabel] = cls[0].Name
	return nil
}

As a result, the deployment is not labelled and thus the corresponding syncer cannot detect the creation.

[gyliu513]

Why did the clustername in the deployment is always admin?

Guangyas-MacBook-Pro:kcp guangyaliu$ kubectl get deploy -oyaml
apiVersion: v1
items:
- apiVersion: apps/v1
  kind: Deployment
  metadata:
    annotations:
      kubectl.kubernetes.io/last-applied-configuration: |
        {"apiVersion":"apps/v1","kind":"Deployment","metadata":{"annotations":{},"name":"example","namespace":"default"},"spec":{"replicas":3,"selector":{"matchLabels":{"app":"nginx"}},"template":{"metadata":{"labels":{"app":"nginx"}},"spec":{"containers":[{"command":["/bin/sh","-ec","echo \"Going to sleep\"\ntail -f /dev/null \n"],"image":"busybox:1.25","name":"busybox"}]}}}}
    clusterName: admin
    ...

[gyliu513]

Even if I change the cluster name to admin, the deployment still cannot sync to downstream cluster.

Guangyas-MacBook-Pro:kcp guangyaliu$ kubectl get cluster
NAME    AGE
admin   86s

[gyliu513]

@y-taka-23 As a simple fix, how about just remove the checking of if len(cls) == 1,