Hmm, fierce fighting. I'll take a look.

cc @davidopp

Since I'm working on refactoring replication controller to use generic etcd, is this something we want for v1?

At least a best-effort check would be nice, but if it's hard we should punt it to post-1.0. It can't be done in validation, unfortunately. See #2395 and #2457. @lavalamp had thoughts about how/where it should be done. I could live without the background checking for 1.0.

related #11990 to clarify the current state of things

fyi #35298

I think overlapping controllers is an obsolete topic with the introduction of owner references in 1.6.

My understanding is that ControllerRef was about protecting the system if the user creates overlapping controllers, but we still don't want the user to do that. If that's the case, we may still want to consider some mechanism for discouraging it.

@enisoc seems kubectl still only uses selector to find matching pods, result in confusing outputs

$ kubectl describe rs replicaset-2
Name:           replicaset-2      
Namespace:      default
Selector:       environment=prod
Labels:         environment=prod
Annotations:    <none>
Replicas:       2 current / 2 desired
Pods Status:    3 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:       environment=prod
  Containers:
   created-from-replicaset:
    Image:              nginx
    Port:               
    Environment:        <none>
    Mounts:             <none>
  Volumes:              <none>
Events:
  FirstSeen     LastSeen        Count   From                    SubObjectPath   Type            Reason                  Message
  ---------     --------        -----   ----                    -------------   --------        ------                  -------
  5m            5m              1       replicaset-controller                   Normal          SuccessfulCreate        Created pod: replicaset-2-39szb
  5m            5m              1       replicaset-controller                   Normal          SuccessfulCreate        Created pod: replicaset-2-470jr

It reports three running pods (another one being managed via another controller with the same selector), should we add more option while listing?

options := metav1.ListOptions{LabelSelector: selector.String()}

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

Prevent issues from auto-closing with an /lifecycle frozen comment.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or @fejta.
/lifecycle stale

ownerReferences is the solution to this.