Merge pull request #124933 from AxeZhan/fix_panic

[Scheduler] Use allNodes when calculating nextStartNodeIndex

pkg/scheduler/schedule_one.go

@@ -500,7 +500,7 @@ func (sched *Scheduler) findNodesThatFitPod(ctx context.Context, fwk framework.F
 	// always try to update the sched.nextStartNodeIndex regardless of whether an error has occurred
 	// this is helpful to make sure that all the nodes have a chance to be searched
 	processedNodes := len(feasibleNodes) + len(diagnosis.NodeToStatusMap)
-	sched.nextStartNodeIndex = (sched.nextStartNodeIndex + processedNodes) % len(nodes)
+	sched.nextStartNodeIndex = (sched.nextStartNodeIndex + processedNodes) % len(allNodes)
 	if err != nil {
 		return nil, diagnosis, err
 	}

pkg/scheduler/schedule_one_test.go

@@ -2433,6 +2433,33 @@ func TestSchedulerSchedulePod(t *testing.T) {
 			// since this case has no score plugin, we'll only try to find one node in Filter stage
 			wantEvaluatedNodes: ptr.To[int32](1),
 		},
+		{
+			name: "test prefilter plugin returned an invalid node",
+			registerPlugins: []tf.RegisterPluginFunc{
+				tf.RegisterQueueSortPlugin(queuesort.Name, queuesort.New),
+				tf.RegisterPreFilterPlugin(
+					"FakePreFilter",
+					tf.NewFakePreFilterPlugin("FakePreFilter", &framework.PreFilterResult{
+						NodeNames: sets.New("invalid-node"),
+					}, nil),
+				),
+				tf.RegisterFilterPlugin("TrueFilter", tf.NewTrueFilterPlugin),
+				tf.RegisterBindPlugin(defaultbinder.Name, defaultbinder.New),
+			},
+			nodes:     []string{"1", "2"},
+			pod:       st.MakePod().Name("test-prefilter").UID("test-prefilter").Obj(),
+			wantNodes: nil,
+			wErr: &framework.FitError{
+				Pod:         st.MakePod().Name("test-prefilter").UID("test-prefilter").Obj(),
+				NumAllNodes: 2,
+				Diagnosis: framework.Diagnosis{
+					NodeToStatusMap: framework.NodeToStatusMap{
+						"1": framework.NewStatus(framework.UnschedulableAndUnresolvable, "node is filtered out by the prefilter result"),
+						"2": framework.NewStatus(framework.UnschedulableAndUnresolvable, "node is filtered out by the prefilter result"),
+					},
+				},
+			},
+		},
 	}
 	for _, test := range tests {
 		t.Run(test.name, func(t *testing.T) {

test/integration/scheduler/scheduler_test.go

@@ -480,6 +480,33 @@ func TestSchedulerInformers(t *testing.T) {
 			}),
 			preemptedPodIndexes: map[int]struct{}{2: {}},
 		},
+		{
+			name:         "The pod cannot be scheduled when nodeAffinity specifies a non-existent node.",
+			nodes:        []*nodeConfig{{name: "node-1", res: defaultNodeRes}},
+			existingPods: []*v1.Pod{},
+			pod: testutils.InitPausePod(&testutils.PausePodConfig{
+				Name:      "unschedulable-pod",
+				Namespace: testCtx.NS.Name,
+				Affinity: &v1.Affinity{
+					NodeAffinity: &v1.NodeAffinity{
+						RequiredDuringSchedulingIgnoredDuringExecution: &v1.NodeSelector{
+							NodeSelectorTerms: []v1.NodeSelectorTerm{
+								{
+									MatchFields: []v1.NodeSelectorRequirement{
+										{
+											Key:      "metadata.name",
+											Operator: v1.NodeSelectorOpIn,
+											Values:   []string{"invalid-node"},
+										},
+									},
+								},
+							},
+						},
+					},
+				},
+				Resources: defaultPodRes,
+			}),
+		},
 	}
 
 	for _, test := range tests {