Add RayClusterProvisioned Condition Type (#2301)

Author: Yicheng-Lu-llll

ray-operator/apis/ray/v1/raycluster_types.go

@@ -168,17 +168,17 @@ type RayClusterConditionType string
 
 // Custom Reason for RayClusterCondition
 const (
-	AllPodRunningAndReady  = "AllPodRunningAndReady"
-	HeadPodNotFound        = "HeadPodNotFound"
-	HeadPodRunningAndReady = "HeadPodRunningAndReady"
+	AllPodRunningAndReadyFirstTime = "AllPodRunningAndReadyFirstTime"
+	HeadPodNotFound                = "HeadPodNotFound"
+	HeadPodRunningAndReady         = "HeadPodRunningAndReady"
 	// UnknownReason says that the reason for the condition is unknown.
 	UnknownReason = "Unknown"
 )
 
 const (
-	// RayClusterReady indicates whether all Ray Pods are ready when the RayCluster is first created.
-	// After RayClusterReady is set to true for the first time, it only indicates whether the RayCluster's head Pod is ready for requests.
-	RayClusterReady RayClusterConditionType = "RayClusterReady"
+	// RayClusterProvisioned indicates whether all Ray Pods are ready for the first time.
+	// After RayClusterProvisioned is set to true for the first time, it will not change anymore.
+	RayClusterProvisioned RayClusterConditionType = "RayClusterProvisioned"
 	// HeadPodReady indicates whether RayCluster's head Pod is ready for requests.
 	HeadPodReady RayClusterConditionType = "HeadPodReady"
 	// RayClusterReplicaFailure is added in a RayCluster when one of its pods fails to be created or deleted.

ray-operator/controllers/ray/raycluster_controller.go

@@ -1210,25 +1210,17 @@ func (r *RayClusterReconciler) calculateStatus(ctx context.Context, instance *ra
 			meta.SetStatusCondition(&newInstance.Status.Conditions, headPodReadyCondition)
 		}
 
-		if meta.FindStatusCondition(newInstance.Status.Conditions, string(rayv1.RayClusterReady)) == nil {
-			// RayClusterReady indicates whether all Ray Pods are ready when the RayCluster is first created.
-			// Note RayClusterReady StatusCondition will not be added to Raycluster until all Ray Pods are ready for the first time.
+		if meta.FindStatusCondition(newInstance.Status.Conditions, string(rayv1.RayClusterProvisioned)) == nil {
+			// RayClusterProvisioned indicates whether all Ray Pods are ready when the RayCluster is first created.
+			// Note RayClusterProvisioned StatusCondition will not be added to Raycluster until all Ray Pods are ready for the first time.
 			if utils.CheckAllPodsRunning(ctx, runtimePods) {
 				meta.SetStatusCondition(&newInstance.Status.Conditions, metav1.Condition{
-					Type:    string(rayv1.RayClusterReady),
+					Type:    string(rayv1.RayClusterProvisioned),
 					Status:  metav1.ConditionTrue,
-					Reason:  rayv1.AllPodRunningAndReady,
-					Message: "All Ray Pods are ready. Future checks focus on the head",
+					Reason:  rayv1.AllPodRunningAndReadyFirstTime,
+					Message: "All Ray Pods are ready for the first time",
 				})
 			}
-		} else { // After RayClusterReady is set to true for the first time, its meaning changes to be the same as HeadPodReady.
-			headPodReadyCondition := meta.FindStatusCondition(newInstance.Status.Conditions, string(rayv1.HeadPodReady))
-			meta.SetStatusCondition(&newInstance.Status.Conditions, metav1.Condition{
-				Type:    string(rayv1.RayClusterReady),
-				Status:  headPodReadyCondition.Status,
-				Reason:  headPodReadyCondition.Reason,
-				Message: "Only check head after all Ray Pods are initially ready",
-			})
 		}
 
 	}

ray-operator/controllers/ray/raycluster_controller_test.go

@@ -834,7 +834,7 @@ var _ = Context("Inside the default namespace", func() {
 			DeferCleanup(cleanUpFunc)
 		})
 
-		It("Should handle HeadPodReady and RayClusterReady conditions correctly", func(ctx SpecContext) {
+		It("Should handle HeadPodReady and RayClusterProvisioned conditions correctly", func(ctx SpecContext) {
 			namespace := "default"
 			rayCluster := rayClusterTemplate("raycluster-status-conditions-enabled", namespace)
 			rayCluster.Spec.WorkerGroupSpecs[0].Replicas = ptr.To[int32](1)
@@ -873,7 +873,7 @@ var _ = Context("Inside the default namespace", func() {
 			headPod = headPods.Items[0]
 
 			By("Check RayCluster conditions empty initially")
-			// Initially, neither head Pod nor worker Pod are ready. The RayClusterReady condition should not be present.
+			// Initially, neither head Pod nor worker Pod are ready. The RayClusterProvisioned condition should not be present.
 			Expect(testRayCluster.Status.Conditions).To(BeEmpty())
 
 			By("Update the head pod to Running and Ready")
@@ -887,7 +887,7 @@ var _ = Context("Inside the default namespace", func() {
 			Expect(k8sClient.Status().Update(ctx, &headPod)).Should(Succeed())
 
 			By("Check RayCluster HeadPodReady condition is true")
-			// The head pod is ready, so RayClusterReady condition should be added and set to True.
+			// The head pod is ready, so HeadPodReady condition should be added and set to True.
 			Eventually(
 				func() bool {
 					if err := getResourceFunc(ctx, client.ObjectKey{Name: rayCluster.Name, Namespace: namespace}, rayCluster)(); err != nil {
@@ -897,14 +897,14 @@ var _ = Context("Inside the default namespace", func() {
 				},
 				time.Second*3, time.Millisecond*500).Should(BeTrue())
 
-			By("Check RayCluster RayClusterReady condition is false")
-			// But the worker pod is not ready yet, RayClusterReady condition should still be absent.
+			By("Check RayCluster RayClusterProvisioned condition is false")
+			// But the worker pod is not ready yet, RayClusterProvisioned condition should still be absent.
 			Consistently(
 				func() *metav1.Condition {
 					if err := getResourceFunc(ctx, client.ObjectKey{Name: rayCluster.Name, Namespace: namespace}, rayCluster)(); err != nil {
 						return nil
 					}
-					return meta.FindStatusCondition(rayCluster.Status.Conditions, string(rayv1.RayClusterReady))
+					return meta.FindStatusCondition(rayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned))
 				},
 				time.Second*3, time.Millisecond*500).Should(BeNil())
 
@@ -918,14 +918,14 @@ var _ = Context("Inside the default namespace", func() {
 			}
 			Expect(k8sClient.Status().Update(ctx, &workerPod)).Should(Succeed())
 
-			By("Check RayCluster RayClusterReady condition is true")
-			// All Ray Pods are ready for the first time, RayClusterReady condition should be added and set to True.
+			By("Check RayCluster RayClusterProvisioned condition is true")
+			// All Ray Pods are ready for the first time, RayClusterProvisioned condition should be added and set to True.
 			Eventually(
 				func() bool {
 					if err := getResourceFunc(ctx, client.ObjectKey{Name: rayCluster.Name, Namespace: namespace}, rayCluster)(); err != nil {
 						return false
 					}
-					return meta.IsStatusConditionPresentAndEqual(rayCluster.Status.Conditions, string(rayv1.RayClusterReady), metav1.ConditionTrue)
+					return meta.IsStatusConditionPresentAndEqual(rayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned), metav1.ConditionTrue)
 				},
 				time.Second*3, time.Millisecond*500).Should(BeTrue())
 
@@ -938,15 +938,15 @@ var _ = Context("Inside the default namespace", func() {
 			}
 			Expect(k8sClient.Status().Update(ctx, &workerPod)).Should(Succeed())
 
-			By("Check RayCluster RayClusterReady condition is true")
-			// The worker pod fails readiness, but since RayClusterReady focuses solely on the headPod after all Ray Pods were initially ready,
-			// RayClusterReady condition should still be True.
+			By("Check RayCluster RayClusterProvisioned condition is true")
+			// The worker pod fails readiness, but since RayClusterProvisioned focuses solely on whether all Ray Pods are ready for the first time,
+			// RayClusterProvisioned condition should still be True.
 			Consistently(
 				func() bool {
 					if err := getResourceFunc(ctx, client.ObjectKey{Name: rayCluster.Name, Namespace: namespace}, rayCluster)(); err != nil {
 						return false
 					}
-					return meta.IsStatusConditionPresentAndEqual(rayCluster.Status.Conditions, string(rayv1.RayClusterReady), metav1.ConditionTrue)
+					return meta.IsStatusConditionPresentAndEqual(rayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned), metav1.ConditionTrue)
 				},
 				time.Second*3, time.Millisecond*500).Should(BeTrue())
 
@@ -970,14 +970,14 @@ var _ = Context("Inside the default namespace", func() {
 				},
 				time.Second*3, time.Millisecond*500).Should(BeTrue())
 
-			By("Check RayCluster RayClusterReady condition is false")
-			// The head pod also fails readiness, RayClusterReady condition should set to False.
+			By("Check RayCluster RayClusterProvisioned condition is still true")
+			// The head pod also fails readiness, RayClusterProvisioned condition not changed.
 			Eventually(
 				func() bool {
 					if err := getResourceFunc(ctx, client.ObjectKey{Name: rayCluster.Name, Namespace: namespace}, rayCluster)(); err != nil {
 						return false
 					}
-					return meta.IsStatusConditionPresentAndEqual(rayCluster.Status.Conditions, string(rayv1.RayClusterReady), metav1.ConditionFalse)
+					return meta.IsStatusConditionPresentAndEqual(rayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned), metav1.ConditionTrue)
 				},
 				time.Second*3, time.Millisecond*500).Should(BeTrue())
 		})

ray-operator/controllers/ray/raycluster_controller_unit_test.go

@@ -1748,7 +1748,7 @@ func TestCalculateStatus(t *testing.T) {
 	assert.True(t, meta.IsStatusConditionPresentAndEqual(newInstance.Status.Conditions, string(rayv1.RayClusterReplicaFailure), metav1.ConditionTrue))
 }
 
-func TestRayClusterReadyCondition(t *testing.T) {
+func TestRayClusterProvisionedCondition(t *testing.T) {
 	setupTest(t)
 	defer features.SetFeatureGateDuringTest(t, features.RayClusterStatusConditions, true)()
 
@@ -1808,40 +1808,40 @@ func TestRayClusterReadyCondition(t *testing.T) {
 		Scheme:   scheme.Scheme,
 	}
 
-	// Initially, neither head Pod nor worker Pod are ready. The RayClusterReady condition should not be present.
+	// Initially, neither head Pod nor worker Pod are ready. The RayClusterProvisioned condition should not be present.
 	headPod.Status = UnReadyStatus
 	workerPod.Status = UnReadyStatus
 	_ = fakeClient.Status().Update(ctx, headPod)
 	_ = fakeClient.Status().Update(ctx, workerPod)
 	testRayCluster, _ = r.calculateStatus(ctx, testRayCluster, nil)
-	assert.Nil(t, meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterReady)))
+	assert.Nil(t, meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned)))
 
-	// After a while, all Ray Pods are ready for the first time, RayClusterReady condition should be added and set to True.
+	// After a while, all Ray Pods are ready for the first time, RayClusterProvisioned condition should be added and set to True.
 	headPod.Status = ReadyStatus
 	workerPod.Status = ReadyStatus
 	_ = fakeClient.Status().Update(ctx, headPod)
 	_ = fakeClient.Status().Update(ctx, workerPod)
 	testRayCluster, _ = r.calculateStatus(ctx, testRayCluster, nil)
-	rayClusterReadyCondition := meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterReady))
-	assert.Equal(t, rayClusterReadyCondition.Status, metav1.ConditionTrue)
-	assert.Equal(t, rayClusterReadyCondition.Reason, rayv1.AllPodRunningAndReady)
+	rayClusterProvisionedCondition := meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned))
+	assert.Equal(t, rayClusterProvisionedCondition.Status, metav1.ConditionTrue)
+	assert.Equal(t, rayClusterProvisionedCondition.Reason, rayv1.AllPodRunningAndReadyFirstTime)
 
-	// After a while, worker Pod fails readiness, but since RayClusterReady focuses solely on the headPod after all Ray Pods were initially ready,
-	// RayClusterReady condition should still be True.
+	// After a while, worker Pod fails readiness, but since RayClusterProvisioned focuses solely on whether all Ray Pods are ready for the first time,
+	// RayClusterProvisioned condition should still be True.
 	workerPod.Status = UnReadyStatus
 	_ = fakeClient.Status().Update(ctx, workerPod)
 	testRayCluster, _ = r.calculateStatus(ctx, testRayCluster, nil)
-	rayClusterReadyCondition = meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterReady))
-	assert.Equal(t, rayClusterReadyCondition.Status, metav1.ConditionTrue)
-	assert.Equal(t, rayClusterReadyCondition.Reason, rayv1.HeadPodRunningAndReady)
+	rayClusterProvisionedCondition = meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned))
+	assert.Equal(t, rayClusterProvisionedCondition.Status, metav1.ConditionTrue)
+	assert.Equal(t, rayClusterProvisionedCondition.Reason, rayv1.AllPodRunningAndReadyFirstTime)
 
-	// After a while, head Pod also fails readiness, RayClusterReady condition should set to False.
+	// After a while, head Pod also fails readiness, RayClusterProvisioned condition should still be true.
 	headPod.Status = UnReadyStatus
 	_ = fakeClient.Status().Update(ctx, headPod)
 	testRayCluster, _ = r.calculateStatus(ctx, testRayCluster, nil)
-	rayClusterReadyCondition = meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterReady))
-	assert.Equal(t, rayClusterReadyCondition.Status, metav1.ConditionFalse)
-	assert.Equal(t, rayClusterReadyCondition.Reason, rayv1.UnknownReason)
+	rayClusterProvisionedCondition = meta.FindStatusCondition(testRayCluster.Status.Conditions, string(rayv1.RayClusterProvisioned))
+	assert.Equal(t, rayClusterProvisionedCondition.Status, metav1.ConditionTrue)
+	assert.Equal(t, rayClusterProvisionedCondition.Reason, rayv1.AllPodRunningAndReadyFirstTime)
 }
 
 func TestStateTransitionTimes_NoStateChange(t *testing.T) {