kubernetes/kops · error

failed to create host %s/%s: %w

Error message

failed to create host %s/%s: %w

What it means

After building the controller-runtime client, enrollHost creates the kops Host custom resource in the kops-system namespace on the target cluster. The error wraps the API server's rejection of that Create call, including its reason.

Source

Thrown at pkg/commands/toolbox_enroll.go:235

		return fmt.Errorf("building kubernetes scheme: %w", err)
	}
	// Ensure that we don't try to use proto with our CRD
	restConfigNoProto := rest.CopyConfig(restConfig)
	restConfigNoProto.ContentType = runtime.ContentTypeJSON
	restConfigNoProto.AcceptContentTypes = runtime.ContentTypeJSON

	kubeClient, err := client.New(restConfigNoProto, client.Options{
		Scheme: scheme,
	})
	if err != nil {
		return fmt.Errorf("building kubernetes client: %w", err)
	}

	// We can't create the host resource in the API server for control-plane nodes,
	// because the API server (likely) isn't running yet.
	if !ig.IsControlPlane() {
		if err := kubeClient.Create(ctx, hostData); err != nil {
			return fmt.Errorf("failed to create host %s/%s: %w", hostData.Namespace, hostData.Name, err)
		}
	}

	for k, v := range bootstrapData.NodeupScriptAdditionalFiles {
		if err := sshTarget.writeFile(ctx, k, bytes.NewReader(v)); err != nil {
			return fmt.Errorf("writing file %q over SSH: %w", k, err)
		}
	}

	if len(bootstrapData.NodeupScript) != 0 {
		if _, err := sshTarget.runScript(ctx, string(bootstrapData.NodeupScript), ExecOptions{Echo: true}); err != nil {
			return err
		}
	}
	return nil
}

const scriptCreateKey = `

View on GitHub (pinned to 4c8573c808)

Solutions

1. Check the wrapped error: if AlreadyExists, the host is registered — delete it first (kubectl -n kops-system delete host <name>) or proceed
2. Confirm the Host CRD exists: kubectl get crd hosts.kops.k8s.io; if missing, upgrade/apply kops CRDs
3. Verify RBAC: kubectl auth can-i create hosts.kops.k8s.io -n kops-system
4. Confirm API server connectivity: kubectl get --raw /healthz

Example fix

// before
if err := kubeClient.Create(ctx, hostData); err != nil {
    return fmt.Errorf("failed to create host %s/%s: %w", hostData.Namespace, hostData.Name, err)
}
// after
err = kubeClient.Create(ctx, hostData)
if err != nil {
    if apierrors.IsAlreadyExists(err) {
        klog.Infof("host %s/%s already registered", hostData.Namespace, hostData.Name)
    } else {
        return fmt.Errorf("failed to create host %s/%s: %w", hostData.Namespace, hostData.Name, err)
    }
}

kubectl get crd hosts.kops.k8s.io && \
  kubectl auth can-i create hosts.kops.k8s.io -n kops-system && \
  kubectl get --raw /healthz

func canCreateHosts(cfg *rest.Config) bool {
    cl, err := discovery.NewForConfig(cfg)
    if err != nil { return false }
    res, err := cl.ServerResourcesForGroupVersion("kops.k8s.io/v1alpha2")
    if err != nil { return false }
    for _, r := range res.APIResources { return strings.Contains(r.Name, "hosts") }
    return false
}

err = kubeClient.Create(ctx, hostData)
if err != nil {
    if apierrors.IsAlreadyExists(err) { klog.Info("host already registered") } else { return err }
}

When it happens

Trigger: kubeClient.Create(ctx, hostData) returns an error: Host CRD not installed on the cluster, the Host with that namespace/name already exists (AlreadyExists), RBAC denies creating hosts, or the API server is unreachable.

Common situations: Enrolling a worker against a cluster where kops-controller CRDs are missing or an older kOps version, re-running enroll for a host that already registered, or insufficient user credentials against the admin API server.

Related errors

• error querying namespace %q: %v
• unable to find resource for %s: %w
• listing nodes in cluster: %v
• error listing nodes in cluster: %v
• error listing nodes: %v

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/938d4919d56d7e5d. Report an issue: GitHub.