feat: sync Cluster conditions to MCP

api/core/v2alpha1/constants.go

@@ -28,6 +28,8 @@ const (
 	ConditionMeta = "Meta"
 
 	ConditionClusterRequestReady       = "ClusterRequestReady"
+	ConditionClusterConditionsSynced   = "ClusterConditionsSynced"
+	ConditionPrefixClusterCondition    = "Cluster."
 	ConditionPrefixOIDCAccessReady     = "OIDCAccessReady."
 	ConditionAllAccessReady            = "AllAccessReady"
 	ConditionAllServicesDeleted        = "AllServicesDeleted"

internal/controllers/managedcontrolplane/clusters.go

@@ -17,15 +17,15 @@ import (
 	corev2alpha1 "github.com/openmcp-project/openmcp-operator/api/core/v2alpha1"
 )
 
-func (r *ManagedControlPlaneReconciler) deleteRelatedClusterRequests(ctx context.Context, mcp *corev2alpha1.ManagedControlPlaneV2, platformNamespace string) (sets.Set[string], errutils.ReasonableError) {
+func (r *ManagedControlPlaneReconciler) deleteRelatedClusterRequests(ctx context.Context, mcp *corev2alpha1.ManagedControlPlaneV2, platformNamespace string) (sets.Set[string], *clustersv1alpha1.Cluster, errutils.ReasonableError) {
 	log := logging.FromContextOrPanic(ctx)
 
 	// delete depending cluster requests, if any
 	crNames := sets.New[string]()
 
 	if mcp == nil {
 		log.Debug("MCP is nil, no need to check for cluster requests")
-		return crNames, nil
+		return crNames, nil, nil
 	}
 
 	// identify cluster request finalizers
@@ -37,7 +37,7 @@ func (r *ManagedControlPlaneReconciler) deleteRelatedClusterRequests(ctx context
 
 	if crNames.Len() == 0 {
 		log.Debug("No cluster request finalizers found on MCP")
-		return crNames, nil
+		return crNames, nil, nil
 	}
 
 	// fetch cluster requests, if any exist
@@ -56,18 +56,34 @@ func (r *ManagedControlPlaneReconciler) deleteRelatedClusterRequests(ctx context
 		resources[crName] = cr
 	}
 	if rerr := errs.Aggregate(); rerr != nil {
-		return sets.KeySet(resources), rerr
+		return sets.KeySet(resources), nil, rerr
 	}
 
 	// delete cluster requests
+	var cluster *clustersv1alpha1.Cluster
 	errs = errutils.NewReasonableErrorList()
 	for crName, cr := range resources {
-		if crName == mcp.Name && len(resources) > 1 {
-			// skip the MCP's main ClusterRequest for now
-			// we want to make sure that all other ClusterRequests are deleted first
-			// in case the corresponding clusters are hosting resources that depend on the MCP cluster
-			log.Debug("Skipping deletion of MCP's primary ClusterRequest, because there are other ClusterRequests to delete first", "crName", crName, "namespace", cr.GetNamespace())
-			continue
+		if crName == mcp.Name {
+			// this is the primary ClusterRequest for the MCP cluster
+			// try to fetch the corresponding Cluster resource
+			// to sync its conditions to the MCP
+			if cr != nil && cr.Status.Cluster != nil {
+				cluster = &clustersv1alpha1.Cluster{}
+				cluster.Name = cr.Status.Cluster.Name
+				cluster.Namespace = cr.Status.Cluster.Namespace
+				if err := r.PlatformCluster.Client().Get(ctx, client.ObjectKeyFromObject(cluster), cluster); err != nil {
+					// only log the error, this is not critical and should not break the function
+					log.Error(fmt.Errorf("unable to get Cluster '%s/%s': %w", cluster.Namespace, cluster.Name, err), "error trying to fetch the primary MCP Cluster resource for condition sync")
+					cluster = nil
+				}
+			}
+			if len(resources) > 1 {
+				// skip the MCP's main ClusterRequest for now
+				// we want to make sure that all other ClusterRequests are deleted first
+				// in case the corresponding clusters are hosting resources that depend on the MCP cluster
+				log.Debug("Skipping deletion of MCP's primary ClusterRequest, because there are other ClusterRequests to delete first", "crName", crName, "namespace", cr.GetNamespace())
+				continue
+			}
 		}
 		if !cr.GetDeletionTimestamp().IsZero() {
 			log.Debug("ClusterRequest resource already marked for deletion", "crName", crName, "namespace", cr.GetNamespace())
@@ -84,8 +100,8 @@ func (r *ManagedControlPlaneReconciler) deleteRelatedClusterRequests(ctx context
 		}
 	}
 	if rerr := errs.Aggregate(); rerr != nil {
-		return sets.KeySet(resources), rerr
+		return sets.KeySet(resources), cluster, rerr
 	}
 
-	return sets.KeySet(resources), nil
+	return sets.KeySet(resources), cluster, nil
 }

internal/controllers/managedcontrolplane/controller.go

@@ -242,6 +242,26 @@ func (r *ManagedControlPlaneReconciler) handleCreateOrUpdate(ctx context.Context
 	log.Debug("ClusterRequest is ready", "clusterRequestName", cr.Name, "clusterRequestNamespace", cr.Namespace)
 	createCon(corev2alpha1.ConditionClusterRequestReady, metav1.ConditionTrue, "", "ClusterRequest is ready")
 
+	// fetch Cluster conditions to display them on the MCP
+	cluster := &clustersv1alpha1.Cluster{}
+	if cr.Status.Cluster == nil {
+		// should not happen if the ClusterRequest is granted
+		rr.ReconcileError = errutils.WithReason(fmt.Errorf("ClusterRequest '%s/%s' does not have a ClusterRef set", cr.Namespace, cr.Name), cconst.ReasonInternalError)
+		createCon(corev2alpha1.ConditionClusterConditionsSynced, metav1.ConditionFalse, rr.ReconcileError.Reason(), rr.ReconcileError.Error())
+		return rr
+	}
+	cluster.Name = cr.Status.Cluster.Name
+	cluster.Namespace = cr.Status.Cluster.Namespace
+	if err := r.PlatformCluster.Client().Get(ctx, client.ObjectKeyFromObject(cluster), cluster); err != nil {
+		rr.ReconcileError = errutils.WithReason(fmt.Errorf("unable to get Cluster '%s/%s': %w", cluster.Namespace, cluster.Name, err), cconst.ReasonPlatformClusterInteractionProblem)
+		createCon(corev2alpha1.ConditionClusterConditionsSynced, metav1.ConditionFalse, rr.ReconcileError.Reason(), rr.ReconcileError.Error())
+		return rr
+	}
+	for _, con := range cluster.Status.Conditions {
+		createCon(corev2alpha1.ConditionPrefixClusterCondition+con.Type, con.Status, con.Reason, con.Message)
+	}
+	createCon(corev2alpha1.ConditionClusterConditionsSynced, metav1.ConditionTrue, "", "Cluster conditions have been synced to MCP")
+
 	// manage AccessRequests
 	allAccessReady, removeConditions, rerr := r.manageAccessRequests(ctx, mcp, platformNamespace, cr, createCon)
 	rr.ConditionsToRemove = removeConditions.UnsortedList()
@@ -259,6 +279,7 @@ func (r *ManagedControlPlaneReconciler) handleCreateOrUpdate(ctx context.Context
 	return rr
 }
 
+//nolint:gocyclo
 func (r *ManagedControlPlaneReconciler) handleDelete(ctx context.Context, mcp *corev2alpha1.ManagedControlPlaneV2) ReconcileResult {
 	log := logging.FromContextOrPanic(ctx)
 	log.Info("Handling deletion of ManagedControlPlane resource")
@@ -325,12 +346,28 @@ func (r *ManagedControlPlaneReconciler) handleDelete(ctx context.Context, mcp *c
 	log.Debug("All AccessRequests deleted")
 
 	// delete cluster requests related to this MCP
-	remainingCRs, rerr := r.deleteRelatedClusterRequests(ctx, mcp, platformNamespace)
+	remainingCRs, primaryCluster, rerr := r.deleteRelatedClusterRequests(ctx, mcp, platformNamespace)
 	if rerr != nil {
 		rr.ReconcileError = rerr
 		createCon(corev2alpha1.ConditionAllClusterRequestsDeleted, metav1.ConditionFalse, rr.ReconcileError.Reason(), rr.ReconcileError.Error())
 		return rr
 	}
+	if primaryCluster != nil {
+		// sync Cluster conditions to the MCP
+		for _, con := range primaryCluster.Status.Conditions {
+			createCon(corev2alpha1.ConditionPrefixClusterCondition+con.Type, con.Status, con.Reason, con.Message)
+		}
+		createCon(corev2alpha1.ConditionClusterConditionsSynced, metav1.ConditionTrue, "", "Cluster conditions have been synced to MCP")
+	} else {
+		// since this point is only reached if no error occurred during r.deleteRelatedClusterRequests, we can assume that the primaryCluster is nil because it does not exist
+		for _, con := range mcp.Status.Conditions {
+			// remove all conditions that were synced from the Cluster from the MCP to avoid having unhealthy leftovers
+			if strings.HasPrefix(con.Type, corev2alpha1.ConditionPrefixClusterCondition) {
+				rr.ConditionsToRemove = append(rr.ConditionsToRemove, con.Type)
+			}
+		}
+		createCon(corev2alpha1.ConditionClusterConditionsSynced, metav1.ConditionTrue, "", "Primary Cluster for MCP does not exist anymore")
+	}
 	finalizersToRemove := sets.New(filters.FilterSlice(mcp.Finalizers, func(args ...any) bool {
 		fin, ok := args[0].(string)
 		if !ok {

internal/controllers/managedcontrolplane/controller_test.go

@@ -62,7 +62,7 @@ func defaultTestSetup(testDirPathSegments ...string) (*managedcontrolplane.Manag
 	}
 	envB := testutils.NewComplexEnvironmentBuilder().
 		WithFakeClient(platform, install.InstallOperatorAPIsPlatform(runtime.NewScheme())).
-		WithDynamicObjectsWithStatus(platform, &clustersv1alpha1.ClusterRequest{}, &clustersv1alpha1.AccessRequest{}).
+		WithDynamicObjectsWithStatus(platform, &clustersv1alpha1.ClusterRequest{}, &clustersv1alpha1.AccessRequest{}, &clustersv1alpha1.Cluster{}).
 		WithFakeClient(onboarding, install.InstallOperatorAPIsOnboarding(runtime.NewScheme())).
 		WithReconcilerConstructor(mcpRec, func(clients ...client.Client) reconcile.Reconciler {
 			mcpr, err := managedcontrolplane.NewManagedControlPlaneReconciler(clusters.NewTestClusterFromClient(platform, clients[0]), clusters.NewTestClusterFromClient(onboarding, clients[1]), nil, cfg.ManagedControlPlane)
@@ -131,15 +131,44 @@ var _ = Describe("ManagedControlPlane Controller", func() {
 				WithReason(cconst.ReasonWaitingForClusterRequest)),
 		))
 
-		// fake ClusterRequest ready status
+		// fake ClusterRequest ready status and Cluster resource
 		By("fake: ClusterRequest readiness")
+		cluster := &clustersv1alpha1.Cluster{}
+		cluster.SetName("cluster-01")
+		cluster.SetNamespace(platformNamespace)
+		cluster.Spec.Purposes = []string{rec.Config.MCPClusterPurpose}
+		Expect(env.Client(platform).Create(env.Ctx, cluster)).To(Succeed())
+		cluster.Status.Conditions = []metav1.Condition{
+			{
+				Type:               "TestCondition1",
+				Status:             metav1.ConditionTrue,
+				Reason:             "TestReason",
+				Message:            "This is a test condition",
+				LastTransitionTime: metav1.Now(),
+				ObservedGeneration: 1,
+			},
+			{
+				Type:               "TestCondition2",
+				Status:             metav1.ConditionFalse,
+				Reason:             "TestReason",
+				Message:            "This is another test condition",
+				LastTransitionTime: metav1.Now(),
+				ObservedGeneration: 1,
+			},
+		}
+		Expect(env.Client(platform).Status().Update(env.Ctx, cluster)).To(Succeed())
 		cr.Status.Phase = clustersv1alpha1.REQUEST_GRANTED
+		cr.Status.Cluster = &commonapi.ObjectReference{
+			Name:      cluster.Name,
+			Namespace: cluster.Namespace,
+		}
 		Expect(env.Client(platform).Status().Update(env.Ctx, cr)).To(Succeed())
 
 		// reconcile the MCP again
 		// expected outcome:
 		// - multiple access requests have been created on the platform cluster, one for each configured OIDC provider
 		// - the mcp has conditions that reflect that it is waiting for the access requests (one for each OIDC provider and one overall one)
+		// - the mcp has taken over the conditions from the Cluster resource with a prefix
 		// - the mcp should be requeued with a short requeueAfter duration
 		By("second MCP reconciliation")
 		res = env.ShouldReconcile(mcpRec, testutils.RequestFromObject(mcp))
@@ -154,6 +183,19 @@ var _ = Describe("ManagedControlPlane Controller", func() {
 				WithType(corev2alpha1.ConditionAllAccessReady).
 				WithStatus(metav1.ConditionFalse).
 				WithReason(cconst.ReasonWaitingForAccessRequest)),
+			MatchCondition(TestCondition().
+				WithType(corev2alpha1.ConditionPrefixClusterCondition+"TestCondition1").
+				WithStatus(metav1.ConditionTrue).
+				WithReason("TestReason").
+				WithMessage("This is a test condition")),
+			MatchCondition(TestCondition().
+				WithType(corev2alpha1.ConditionPrefixClusterCondition+"TestCondition2").
+				WithStatus(metav1.ConditionFalse).
+				WithReason("TestReason").
+				WithMessage("This is another test condition")),
+			MatchCondition(TestCondition().
+				WithType(corev2alpha1.ConditionClusterConditionsSynced).
+				WithStatus(metav1.ConditionTrue)),
 		))
 		oidcProviders := []commonapi.OIDCProviderConfig{*rec.Config.DefaultOIDCProvider.DeepCopy()}
 		oidcProviders[0].RoleBindings = mcp.Spec.IAM.RoleBindings
@@ -552,10 +594,12 @@ var _ = Describe("ManagedControlPlane Controller", func() {
 				WithReason(cconst.ReasonWaitingForClusterRequestDeletion)),
 		))
 
-		// remove finalizer from cr
+		// remove finalizer from cr and remove Cluster resource
 		By("fake: removing finalizer from primary ClusterRequest")
 		controllerutil.RemoveFinalizer(cr, "dummy")
 		Expect(env.Client(platform).Update(env.Ctx, cr)).To(Succeed())
+		Expect(env.Client(platform).Delete(env.Ctx, cluster)).To(Succeed())
+		Expect(env.Client(platform).Get(env.Ctx, client.ObjectKeyFromObject(cluster), cluster)).To(MatchError(apierrors.IsNotFound, "IsNotFound"))
 
 		// add finalizer to MCP namespace
 		By("fake: adding finalizer to MCP namespace")
@@ -567,6 +611,7 @@ var _ = Describe("ManagedControlPlane Controller", func() {
 		// expected outcome:
 		// - the MCP namespace has a deletion timestamp
 		// - the MCP has a condition stating that it is waiting for the MCP namespace to be deleted
+		// - the Cluster conditions should have been removed
 		// - the MCP should be requeued with a short requeueAfter duration
 		By("fifth MCP reconciliation after delete")
 		res = env.ShouldReconcile(mcpRec, testutils.RequestFromObject(mcp))
@@ -580,6 +625,14 @@ var _ = Describe("ManagedControlPlane Controller", func() {
 				WithType(corev2alpha1.ConditionMeta).
 				WithStatus(metav1.ConditionFalse).
 				WithReason(cconst.ReasonWaitingForNamespaceDeletion)),
+			MatchCondition(TestCondition().
+				WithType(corev2alpha1.ConditionClusterConditionsSynced).
+				WithStatus(metav1.ConditionTrue)),
+		))
+		Expect(mcp.Status.Conditions).ToNot(ContainElements(
+			MatchFields(IgnoreExtras, Fields{
+				"Type": ContainSubstring(corev2alpha1.ConditionPrefixClusterCondition),
+			}),
 		))
 
 		// remove finalizer from MCP namespace