What causes "can not rollback current replica" in derailed/k9s?

Select an older ReplicaSet (one with 0 replicas / not the revision annotated as current) and retry the rollback.

can not rollback current replica

Error message

can not rollback current replica

What it means

getRSRevision backs the ReplicaSet rollback command (dao/rs.go:88 Rollback). Kubernetes refuses to roll a Deployment back to the revision that is currently active, so k9s checks rs.Status.Replicas: a non-zero count means this ReplicaSet is the live one and the attempt is rejected before contacting the API server. The error tells you to pick an older revision instead.

Solutions

Select an older ReplicaSet (one with 0 replicas / not the revision annotated as current) and retry the rollback.
Confirm with `kubectl rollout history deployment/<name>` which revision is active, and target a previous one.
If you truly need to reapply the current revision, edit/apply the Deployment spec instead of using rollback.

Defensive patterns

Strategy: validation

Validate before calling

func isRollbackable(rs *appsv1.ReplicaSet) bool {
    return rs.Status.Replicas == 0
}

rs, _ := replicaSetDAO.Load(factory, fqn)
if !isRollbackable(rs) {
    // pick an older RS row; current revision cannot be a rollback target
}

Try / catch

if err := rsDAO.Rollback(fqn); err != nil {
    if strings.Contains(err.Error(), "can not rollback current replica") {
        flash.Warnf("Select an older replicaset to roll back")
    } else { /* surface err */ }
}

Prevention

In UI code, only offer :rollback on rows with 0 replicas.
Teach users to check `kubectl rollout history` for valid targets.
Remember rolling back TO the active revision is a no-op by design.

When it happens

Trigger: Running the rollback action (:rollback) on the newest ReplicaSet of a Deployment while it still has running replicas (Status.Replicas > 0), i.e. trying to roll back to the current revision.

Common situations: Selecting the top/active RS row in the replicasets view instead of an older one; a Deployment that just rolled out and the old RS was scaled to 0 so users assume any row is rollback-able; automated scripts invoking Rollback on an arbitrary RS FQN.

Related errors

AI-assisted analysis of derailed/k9s@2d3ccc6ba2 (2026-08-15). Data as JSON: /api/errors/e43a5d27bd1da505. Report an issue: GitHub.

Appendix: source

Thrown at internal/dao/rs.go:62

func (*ReplicaSet) Load(f Factory, path string) (*appsv1.ReplicaSet, error) {
	o, err := f.Get(client.RsGVR, path, true, labels.Everything())
	if err != nil {
		return nil, err
	}

	var rs appsv1.ReplicaSet
	err = runtime.DefaultUnstructuredConverter.FromUnstructured(o.(*unstructured.Unstructured).Object, &rs)
	if err != nil {
		return nil, err
	}

	return &rs, nil
}

func getRSRevision(rs *appsv1.ReplicaSet) (int64, error) {
	revision := rs.Annotations["deployment.kubernetes.io/revision"]
	if rs.Status.Replicas != 0 {
		return 0, errors.New("can not rollback current replica")
	}
	vers, err := strconv.Atoi(revision)
	if err != nil {
		return 0, errors.New("revision conversion failed")
	}

	return int64(vers), nil
}

func controllerInfo(rs *appsv1.ReplicaSet) (name, kind, group string, err error) {
	for _, ref := range rs.OwnerReferences {
		if ref.Controller == nil {
			continue
		}
		group, tokens := ref.APIVersion, strings.Split(ref.APIVersion, "/")
		if len(tokens) == 2 {
			group = tokens[0]
		}

View on GitHub (pinned to 2d3ccc6ba2)