Read userdata from bootstrap secret.

metal-stack · Nov 14, 2024 · ebe0325 · ebe0325
1 parent 2260356
commit ebe0325
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 1 deletion.
diff --git a/api/v1alpha1/metalstackmachine_types.go b/api/v1alpha1/metalstackmachine_types.go
@@ -29,6 +29,7 @@ const (
 	TagInfraMachineID = "metal-stack.infrastructure.cluster.x-k8s.io/machine-id"
 
 	ProviderMachineCreated clusterv1.ConditionType = "MachineCreated"
+	ProviderMachineReady   clusterv1.ConditionType = "MachineReady"
 	ProviderMachineHealthy clusterv1.ConditionType = "MachineHealthy"
 )
 

diff --git a/config/samples/example.yaml b/config/samples/example.yaml
@@ -34,6 +34,13 @@ apiVersion: controlplane.cluster.x-k8s.io/v1beta1
 metadata:
   name: metal-test-controlplane
 spec:
+  kubeadmConfigSpec:
+    format: ignition
+    initConfiguration:
+      nodeRegistration: {}
+    joinConfiguration:
+      controlPlane: {}
+      nodeRegistration: {}
   machineTemplate:
     nodeDrainTimeout: 10m
     infrastructureRef:

diff --git a/internal/controller/metalstackcluster_controller.go b/internal/controller/metalstackcluster_controller.go
@@ -113,6 +113,8 @@ func (r *MetalStackClusterReconciler) Reconcile(ctx context.Context, req ctrl.Re
 		statusErr := reconciler.status()
 		if statusErr != nil {
 			err = errors.Join(err, fmt.Errorf("unable to update status: %w", statusErr))
+		} else if !reconciler.infraCluster.Status.Ready {
+			err = errors.New("cluster is not yet ready, requeueing")
 		}
 	}()
 

diff --git a/internal/controller/metalstackmachine_controller.go b/internal/controller/metalstackmachine_controller.go
@@ -143,6 +143,8 @@ func (r *MetalStackMachineReconciler) Reconcile(ctx context.Context, req ctrl.Re
 		statusErr := reconciler.status()
 		if statusErr != nil {
 			err = errors.Join(err, fmt.Errorf("unable to update status: %w", statusErr))
+		} else if !reconciler.infraMachine.Status.Ready {
+			err = errors.New("machine is not yet ready, requeueing")
 		}
 	}()
 
@@ -188,6 +190,10 @@ func (r *MetalStackMachineReconciler) Reconcile(ctx context.Context, req ctrl.Re
 		return ctrl.Result{}, errors.New("waiting until control plane ip was set to cluster spec")
 	}
 
+	if machine.Spec.Bootstrap.DataSecretName == nil {
+		return ctrl.Result{}, errors.New("waiting until bootstrap data secret was created")
+	}
+
 	err = reconciler.reconcile()
 
 	return ctrl.Result{}, err // remember to return err here and not nil because the defer func can influence this
@@ -256,6 +262,17 @@ func (r *machineReconciler) delete() error {
 }
 
 func (r *machineReconciler) create() (*models.V1MachineResponse, error) {
+	bootstrapSecret := &corev1.Secret{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      *r.clusterMachine.Spec.Bootstrap.DataSecretName,
+			Namespace: r.infraMachine.Namespace,
+		},
+	}
+	err := r.client.Get(r.ctx, client.ObjectKeyFromObject(bootstrapSecret), bootstrapSecret)
+	if err != nil {
+		return nil, fmt.Errorf("unable to fetch bootstrap secret: %w", err)
+	}
+
 	var (
 		ips []string
 		nws = []*models.V1MachineAllocationNetwork{
@@ -292,7 +309,8 @@ func (r *machineReconciler) create() (*models.V1MachineResponse, error) {
 		Description:   fmt.Sprintf("%s/%s for cluster %s/%s", r.infraMachine.Namespace, r.infraMachine.Name, r.infraCluster.Namespace, r.infraCluster.Name),
 		Networks:      nws,
 		Ips:           ips,
-		// TODO: UserData, SSHPubKeys, ...
+		UserData:      string(bootstrapSecret.Data["value"]),
+		// TODO: SSHPubKeys, ...
 	}), nil)
 	if err != nil {
 		return nil, fmt.Errorf("failed to allocate machine: %w", err)
@@ -330,9 +348,11 @@ func (r *machineReconciler) status() error {
 			case err != nil && !errors.Is(err, errProviderMachineNotFound):
 				conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineCreated, "InternalError", clusterv1.ConditionSeverityError, "%s", err.Error())
 				conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineHealthy, "NotHealthy", clusterv1.ConditionSeverityWarning, "machine not created")
+				conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineReady, "NotReady", clusterv1.ConditionSeverityWarning, "machine not created")
 			case err != nil && errors.Is(err, errProviderMachineNotFound):
 				conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineCreated, "NotCreated", clusterv1.ConditionSeverityError, "%s", err.Error())
 				conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineHealthy, "NotHealthy", clusterv1.ConditionSeverityWarning, "machine not created")
+				conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineReady, "NotReady", clusterv1.ConditionSeverityWarning, "machine not created")
 			default:
 				if r.infraMachine.Spec.ProviderID == *m.ID {
 					conditions.MarkTrue(r.infraMachine, v1alpha1.ProviderMachineCreated)
@@ -357,6 +377,12 @@ func (r *machineReconciler) status() error {
 					}
 				}
 
+				if m.Events != nil && len(m.Events.Log) > 0 && ptr.Deref(m.Events.Log[0].Event, "") == "Phoned Home" {
+					conditions.MarkTrue(r.infraMachine, v1alpha1.ProviderMachineReady)
+				} else {
+					conditions.MarkFalse(r.infraMachine, v1alpha1.ProviderMachineReady, "NotReady", clusterv1.ConditionSeverityWarning, "machine is not in phoned home state")
+				}
+
 				if len(errs) == 0 {
 					conditions.MarkTrue(r.infraMachine, v1alpha1.ProviderMachineHealthy)
 				} else {