Merge pull request #81081 from p0lyn0mial/gc_transition_delete

changes the way `gc` detect that an object needs to be deleted

pkg/controller/garbagecollector/graph_builder.go

@@ -427,15 +427,14 @@ func referencesDiffs(old []metav1.OwnerReference, new []metav1.OwnerReference) (
 	return added, removed, changed
 }
 
-// returns if the object in the event just transitions to "being deleted".
-func deletionStarts(oldObj interface{}, newAccessor metav1.Object) bool {
-	// The delta_fifo may combine the creation and update of the object into one
-	// event, so if there is no oldObj, we just return if the newObj (via
-	// newAccessor) is being deleted.
+func deletionStartsWithFinalizer(oldObj interface{}, newAccessor metav1.Object, matchingFinalizer string) bool {
+	// if the new object isn't being deleted, or doesn't have the finalizer we're interested in, return false
+	if !beingDeleted(newAccessor) || !hasFinalizer(newAccessor, matchingFinalizer) {
+		return false
+	}
+
+	// if the old object is nil, or wasn't being deleted, or didn't have the finalizer, return true
 	if oldObj == nil {
-		if newAccessor.GetDeletionTimestamp() == nil {
-			return false
-		}
 		return true
 	}
 	oldAccessor, err := meta.Accessor(oldObj)
@@ -443,7 +442,7 @@ func deletionStarts(oldObj interface{}, newAccessor metav1.Object) bool {
 		utilruntime.HandleError(fmt.Errorf("cannot access oldObj: %v", err))
 		return false
 	}
-	return beingDeleted(newAccessor) && !beingDeleted(oldAccessor)
+	return !beingDeleted(oldAccessor) || !hasFinalizer(oldAccessor, matchingFinalizer)
 }
 
 func beingDeleted(accessor metav1.Object) bool {
@@ -471,13 +470,13 @@ func hasFinalizer(accessor metav1.Object, matchingFinalizer string) bool {
 // this function takes newAccessor directly because the caller already
 // instantiates an accessor for the newObj.
 func startsWaitingForDependentsDeleted(oldObj interface{}, newAccessor metav1.Object) bool {
-	return deletionStarts(oldObj, newAccessor) && hasDeleteDependentsFinalizer(newAccessor)
+	return deletionStartsWithFinalizer(oldObj, newAccessor, metav1.FinalizerDeleteDependents)
 }
 
 // this function takes newAccessor directly because the caller already
 // instantiates an accessor for the newObj.
 func startsWaitingForDependentsOrphaned(oldObj interface{}, newAccessor metav1.Object) bool {
-	return deletionStarts(oldObj, newAccessor) && hasOrphanFinalizer(newAccessor)
+	return deletionStartsWithFinalizer(oldObj, newAccessor, metav1.FinalizerOrphanDependents)
 }
 
 // if an blocking ownerReference points to an object gets removed, or gets set to

test/integration/garbagecollector/garbage_collector_test.go

@@ -786,6 +786,69 @@ func TestNonBlockingOwnerRefDoesNotBlock(t *testing.T) {
 	}
 }
 
+func TestDoubleDeletionWithFinalizer(t *testing.T) {
+	// test setup
+	ctx := setup(t, 5)
+	defer ctx.tearDown()
+	clientSet := ctx.clientSet
+	ns := createNamespaceOrDie("gc-double-foreground", clientSet, t)
+	defer deleteNamespaceOrDie(ns.Name, clientSet, t)
+
+	// step 1: creates a pod with a custom finalizer and deletes it, then waits until gc removes its finalizer
+	podClient := clientSet.CoreV1().Pods(ns.Name)
+	pod := newPod("lucy", ns.Name, nil)
+	pod.ObjectMeta.Finalizers = []string{"x/y"}
+	if _, err := podClient.Create(pod); err != nil {
+		t.Fatalf("Failed to create pod: %v", err)
+	}
+	if err := podClient.Delete(pod.Name, getForegroundOptions()); err != nil {
+		t.Fatalf("Failed to delete pod: %v", err)
+	}
+	if err := wait.PollImmediate(1*time.Second, 10*time.Second, func() (bool, error) {
+		returnedPod, err := podClient.Get(pod.Name, metav1.GetOptions{})
+		if err != nil {
+			return false, err
+		}
+		if len(returnedPod.Finalizers) != 1 || returnedPod.Finalizers[0] != "x/y" {
+			t.Logf("waiting for pod %q to have only one finalizer %q at step 1, got %v", returnedPod.Name, "x/y", returnedPod.Finalizers)
+			return false, nil
+		}
+		return true, nil
+	}); err != nil {
+		t.Fatalf("Failed waiting for pod to have only one filanizer at step 1, error: %v", err)
+	}
+
+	// step 2: deletes the pod one more time and checks if there's only the custom finalizer left
+	if err := podClient.Delete(pod.Name, getForegroundOptions()); err != nil {
+		t.Fatalf("Failed to delete pod: %v", err)
+	}
+	if err := wait.PollImmediate(1*time.Second, 10*time.Second, func() (bool, error) {
+		returnedPod, err := podClient.Get(pod.Name, metav1.GetOptions{})
+		if err != nil {
+			return false, err
+		}
+		if len(returnedPod.Finalizers) != 1 || returnedPod.Finalizers[0] != "x/y" {
+			t.Logf("waiting for pod %q to have only one finalizer %q at step 2, got %v", returnedPod.Name, "x/y", returnedPod.Finalizers)
+			return false, nil
+		}
+		return true, nil
+	}); err != nil {
+		t.Fatalf("Failed waiting for pod to have only one finalizer at step 2, gc hasn't removed its finalzier?, error: %v", err)
+	}
+
+	// step 3: removes the custom finalizer and checks if the pod was removed
+	patch := []byte(`[{"op":"remove","path":"/metadata/finalizers"}]`)
+	if _, err := podClient.Patch(pod.Name, types.JSONPatchType, patch); err != nil {
+		t.Fatalf("Failed to update pod: %v", err)
+	}
+	if err := wait.Poll(1*time.Second, 10*time.Second, func() (bool, error) {
+		_, err := podClient.Get(pod.Name, metav1.GetOptions{})
+		return errors.IsNotFound(err), nil
+	}); err != nil {
+		t.Fatalf("Failed waiting for pod %q to be deleted", pod.Name)
+	}
+}
+
 func TestBlockingOwnerRefDoesBlock(t *testing.T) {
 	ctx := setup(t, 0)
 	defer ctx.tearDown()