Merge pull request #376 from red-hat-storage/sync_ds--main

Syncing latest changes from main for ramen
red-hat-storage · Oct 15, 2024 · 7802f71 · 7802f71
2 parents d0ccfa5 + e125fe5
commit 7802f71
Show file tree

Hide file tree

Showing 16 changed files with 365 additions and 113 deletions.
diff --git a/internal/controller/drplacementcontrol.go b/internal/controller/drplacementcontrol.go
@@ -825,6 +825,14 @@ func (d *DRPCInstance) RunRelocate() (bool, error) {
 
 	const done = true
 
+	if d.reconciler.numClustersQueriedSuccessfully != len(d.drPolicy.Spec.DRClusters) {
+		d.log.Info("Can't progress with relocation -- Not all clusters are reachable",
+			"numClustersQueriedSuccessfully", d.reconciler.numClustersQueriedSuccessfully,
+			"NumOfClusters", len(d.drPolicy.Spec.DRClusters))
+
+		return !done, nil
+	}
+
 	preferredCluster := d.instance.Spec.PreferredCluster
 	preferredClusterNamespace := d.instance.Spec.PreferredCluster
 

diff --git a/internal/controller/drplacementcontrol_controller.go b/internal/controller/drplacementcontrol_controller.go
@@ -68,15 +68,16 @@ type ProgressCallback func(string, string)
 // DRPlacementControlReconciler reconciles a DRPlacementControl object
 type DRPlacementControlReconciler struct {
 	client.Client
-	APIReader           client.Reader
-	Log                 logr.Logger
-	MCVGetter           rmnutil.ManagedClusterViewGetter
-	Scheme              *runtime.Scheme
-	Callback            ProgressCallback
-	eventRecorder       *rmnutil.EventReporter
-	savedInstanceStatus rmn.DRPlacementControlStatus
-	ObjStoreGetter      ObjectStoreGetter
-	RateLimiter         *workqueue.TypedRateLimiter[reconcile.Request]
+	APIReader                      client.Reader
+	Log                            logr.Logger
+	MCVGetter                      rmnutil.ManagedClusterViewGetter
+	Scheme                         *runtime.Scheme
+	Callback                       ProgressCallback
+	eventRecorder                  *rmnutil.EventReporter
+	savedInstanceStatus            rmn.DRPlacementControlStatus
+	ObjStoreGetter                 ObjectStoreGetter
+	RateLimiter                    *workqueue.TypedRateLimiter[reconcile.Request]
+	numClustersQueriedSuccessfully int
 }
 
 // SetupWithManager sets up the controller with the Manager.
@@ -373,11 +374,13 @@ func (r *DRPlacementControlReconciler) createDRPCInstance(
 		return nil, err
 	}
 
-	vrgs, _, _, err := getVRGsFromManagedClusters(r.MCVGetter, drpc, drClusters, vrgNamespace, log)
+	vrgs, cqs, _, err := getVRGsFromManagedClusters(r.MCVGetter, drpc, drClusters, vrgNamespace, log)
 	if err != nil {
 		return nil, err
 	}
 
+	r.numClustersQueriedSuccessfully = cqs
+
 	d := &DRPCInstance{
 		reconciler:      r,
 		ctx:             ctx,
@@ -1097,7 +1100,7 @@ func getVRGsFromManagedClusters(
 	annotations[DRPCNameAnnotation] = drpc.Name
 	annotations[DRPCNamespaceAnnotation] = drpc.Namespace
 
-	var clustersQueriedSuccessfully int
+	var numClustersQueriedSuccessfully int
 
 	var failedCluster string
 
@@ -1109,7 +1112,7 @@ func getVRGsFromManagedClusters(
 			// Only NotFound error is accepted
 			if errors.IsNotFound(err) {
 				log.Info(fmt.Sprintf("VRG not found on %q", drCluster.Name))
-				clustersQueriedSuccessfully++
+				numClustersQueriedSuccessfully++
 
 				continue
 			}
@@ -1121,7 +1124,7 @@ func getVRGsFromManagedClusters(
 			continue
 		}
 
-		clustersQueriedSuccessfully++
+		numClustersQueriedSuccessfully++
 
 		if rmnutil.ResourceIsDeleted(drCluster) {
 			log.Info("Skipping VRG on deleted drcluster", "drcluster", drCluster.Name, "vrg", vrg.Name)
@@ -1135,15 +1138,15 @@ func getVRGsFromManagedClusters(
 	}
 
 	// We are done if we successfully queried all drClusters
-	if clustersQueriedSuccessfully == len(drClusters) {
-		return vrgs, clustersQueriedSuccessfully, "", nil
+	if numClustersQueriedSuccessfully == len(drClusters) {
+		return vrgs, numClustersQueriedSuccessfully, "", nil
 	}
 
-	if clustersQueriedSuccessfully == 0 {
+	if numClustersQueriedSuccessfully == 0 {
 		return vrgs, 0, "", fmt.Errorf("failed to retrieve VRGs from clusters")
 	}
 
-	return vrgs, clustersQueriedSuccessfully, failedCluster, nil
+	return vrgs, numClustersQueriedSuccessfully, failedCluster, nil
 }
 
 func (r *DRPlacementControlReconciler) deleteClonedPlacementRule(ctx context.Context,

diff --git a/internal/controller/drplacementcontrol_controller_test.go b/internal/controller/drplacementcontrol_controller_test.go
@@ -2479,8 +2479,8 @@ var _ = Describe("DRPlacementControl Reconciler", func() {
 				clearFakeUserPlacementRuleStatus(UserPlacementRuleName, DefaultDRPCNamespace)
 				clearDRPCStatus()
 				expectedAction := rmn.ActionRelocate
-				expectedPhase := rmn.Relocated
-				exptectedPorgression := rmn.ProgressionCleaningUp
+				expectedPhase := rmn.DRState("")
+				exptectedPorgression := rmn.ProgressionStatus("")
 				verifyDRPCStateAndProgression(expectedAction, expectedPhase, exptectedPorgression)
 
 				// User intervention is required (simulate user intervention)

diff --git a/internal/controller/volsync/vshandler_test.go b/internal/controller/volsync/vshandler_test.go
@@ -1216,14 +1216,12 @@ var _ = Describe("VolSync_Handler", func() {
 				pvc := &corev1.PersistentVolumeClaim{}
 				JustBeforeEach(func() {
 					// Common checks for everything in this context - pvc should be created with correct spec
-					Expect(ensurePVCErr).NotTo(HaveOccurred())
-
 					Eventually(func() error {
 						return k8sClient.Get(ctx, types.NamespacedName{
 							Name:      pvcName,
 							Namespace: testNamespace.GetName(),
 						}, pvc)
-					}, maxWait, interval).Should(Succeed())
+					}, maxWait, interval).Should(Succeed(), fmt.Sprintf("Original error %v", ensurePVCErr))
 
 					Expect(pvc.GetName()).To(Equal(pvcName))
 					Expect(pvc.Spec.AccessModes).To(Equal([]corev1.PersistentVolumeAccessMode{corev1.ReadWriteOnce}))