Moved node condition filter into a predicates.

2025-11-03 19:58:17 +00:00 · 2017-08-23 11:00:42 +08:00
parent a235ba4e49
commit 18dc690c7c
9 changed files with 217 additions and 84 deletions
--- a/plugin/pkg/scheduler/algorithm/predicates/error.go
+++ b/plugin/pkg/scheduler/algorithm/predicates/error.go
@@ -38,6 +38,10 @@ var (
 	ErrNodeUnderMemoryPressure   = newPredicateFailureError("NodeUnderMemoryPressure")
 	ErrNodeUnderDiskPressure     = newPredicateFailureError("NodeUnderDiskPressure")
 	ErrNodeOutOfDisk             = newPredicateFailureError("NodeOutOfDisk")
+	ErrNodeNotReady              = newPredicateFailureError("NodeNotReady")
+	ErrNodeNetworkUnavailable    = newPredicateFailureError("NodeNetworkUnavailable")
+	ErrNodeUnschedulable         = newPredicateFailureError("NodeUnschedulable")
+	ErrNodeUnknownCondition      = newPredicateFailureError("NodeUnknownCondition")
 	ErrVolumeNodeConflict        = newPredicateFailureError("NoVolumeNodeConflict")
 	// ErrFakePredicate is used for test only. The fake predicates returning false also returns error
 	// as ErrFakePredicate.
--- a/plugin/pkg/scheduler/algorithm/predicates/predicates.go
+++ b/plugin/pkg/scheduler/algorithm/predicates/predicates.go
@@ -1301,6 +1301,37 @@ func CheckNodeDiskPressurePredicate(pod *v1.Pod, meta interface{}, nodeInfo *sch
 	return true, nil, nil
 }

+// CheckNodeConditionPredicate checks if a pod can be scheduled on a node reporting out of disk,
+// network unavailable and not ready condition. Only node conditions are accounted in this predicate.
+func CheckNodeConditionPredicate(pod *v1.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
+	reasons := []algorithm.PredicateFailureReason{}
+
+	if nodeInfo == nil || nodeInfo.Node() == nil {
+		return false, []algorithm.PredicateFailureReason{ErrNodeUnknownCondition}, nil
+	}
+
+	node := nodeInfo.Node()
+	for _, cond := range node.Status.Conditions {
+		// We consider the node for scheduling only when its:
+		// - NodeReady condition status is ConditionTrue,
+		// - NodeOutOfDisk condition status is ConditionFalse,
+		// - NodeNetworkUnavailable condition status is ConditionFalse.
+		if cond.Type == v1.NodeReady && cond.Status != v1.ConditionTrue {
+			reasons = append(reasons, ErrNodeNotReady)
+		} else if cond.Type == v1.NodeOutOfDisk && cond.Status != v1.ConditionFalse {
+			reasons = append(reasons, ErrNodeOutOfDisk)
+		} else if cond.Type == v1.NodeNetworkUnavailable && cond.Status != v1.ConditionFalse {
+			reasons = append(reasons, ErrNodeNetworkUnavailable)
+		}
+	}
+
+	if node.Spec.Unschedulable {
+		reasons = append(reasons, ErrNodeUnschedulable)
+	}
+
+	return len(reasons) == 0, reasons, nil
+}
+
 type VolumeNodeChecker struct {
 	pvInfo  PersistentVolumeInfo
 	pvcInfo PersistentVolumeClaimInfo
--- a/plugin/pkg/scheduler/algorithm/predicates/predicates_test.go
+++ b/plugin/pkg/scheduler/algorithm/predicates/predicates_test.go
@@ -3442,6 +3442,78 @@ func TestPodSchedulesOnNodeWithDiskPressureCondition(t *testing.T) {
 	}
 }

+func TestNodeConditionPredicate(t *testing.T) {
+	tests := []struct {
+		node        *v1.Node
+		schedulable bool
+	}{
+		// node1 considered
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node1"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionTrue}}}},
+			schedulable: true,
+		},
+		// node2 ignored - node not Ready
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node2"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionFalse}}}},
+			schedulable: false,
+		},
+		// node3 ignored - node out of disk
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node3"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeOutOfDisk, Status: v1.ConditionTrue}}}},
+			schedulable: false,
+		},
+
+		// node4 considered
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node4"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeOutOfDisk, Status: v1.ConditionFalse}}}},
+			schedulable: true,
+		},
+		// node5 ignored - node out of disk
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node5"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionTrue}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionTrue}}}},
+			schedulable: false,
+		},
+		// node6 considered
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node6"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionTrue}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionFalse}}}},
+			schedulable: true,
+		},
+		// node7 ignored - node out of disk, node not Ready
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node7"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionFalse}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionTrue}}}},
+			schedulable: false,
+		},
+		// node8 ignored - node not Ready
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node8"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionFalse}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionFalse}}}},
+			schedulable: false,
+		},
+		// node9 ignored - node unschedulable
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node9"}, Spec: v1.NodeSpec{Unschedulable: true}},
+			schedulable: false,
+		},
+		// node10 considered
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node10"}, Spec: v1.NodeSpec{Unschedulable: false}},
+			schedulable: true,
+		},
+		// node11 considered
+		{
+			node:        &v1.Node{ObjectMeta: metav1.ObjectMeta{Name: "node11"}},
+			schedulable: true,
+		},
+	}
+
+	for _, test := range tests {
+		nodeInfo := makeEmptyNodeInfo(test.node)
+		if fit, reasons, err := CheckNodeConditionPredicate(nil, nil, nodeInfo); fit != test.schedulable {
+			t.Errorf("%s: expected: %t, got %t; %+v, %v",
+				test.node.Name, test.schedulable, fit, reasons, err)
+		}
+	}
+}
+
 func createPodWithVolume(pod, pv, pvc string) *v1.Pod {
 	return &v1.Pod{
 		ObjectMeta: metav1.ObjectMeta{Name: pod, Namespace: "default"},
--- a/plugin/pkg/scheduler/algorithmprovider/defaults/compatibility_test.go
+++ b/plugin/pkg/scheduler/algorithmprovider/defaults/compatibility_test.go
@@ -387,6 +387,79 @@ func TestCompatibility_v1_Scheduler(t *testing.T) {
 				},
 			},
 		},
+		// Do not change this JSON after the corresponding release has been tagged.
+		// A failure indicates backwards compatibility with the specified release was broken.
+		"1.8": {
+			JSON: `{
+		  "kind": "Policy",
+		  "apiVersion": "v1",
+		  "predicates": [
+			{"name": "MatchNodeSelector"},
+			{"name": "PodFitsResources"},
+			{"name": "PodFitsHostPorts"},
+			{"name": "HostName"},
+			{"name": "NoDiskConflict"},
+			{"name": "NoVolumeZoneConflict"},
+			{"name": "PodToleratesNodeTaints"},
+			{"name": "CheckNodeMemoryPressure"},
+			{"name": "CheckNodeDiskPressure"},
+			{"name": "CheckNodeCondition"},
+			{"name": "MaxEBSVolumeCount"},
+			{"name": "MaxGCEPDVolumeCount"},
+			{"name": "MaxAzureDiskVolumeCount"},
+			{"name": "MatchInterPodAffinity"},
+			{"name": "GeneralPredicates"},
+			{"name": "TestServiceAffinity", "argument": {"serviceAffinity" : {"labels" : ["region"]}}},
+			{"name": "TestLabelsPresence",  "argument": {"labelsPresence"  : {"labels" : ["foo"], "presence":true}}},
+			{"name": "NoVolumeNodeConflict"}
+		  ],"priorities": [
+			{"name": "EqualPriority",   "weight": 2},
+			{"name": "ImageLocalityPriority",   "weight": 2},
+			{"name": "LeastRequestedPriority",   "weight": 2},
+			{"name": "BalancedResourceAllocation",   "weight": 2},
+			{"name": "SelectorSpreadPriority",   "weight": 2},
+			{"name": "NodePreferAvoidPodsPriority",   "weight": 2},
+			{"name": "NodeAffinityPriority",   "weight": 2},
+			{"name": "TaintTolerationPriority",   "weight": 2},
+			{"name": "InterPodAffinityPriority",   "weight": 2},
+			{"name": "MostRequestedPriority",   "weight": 2}
+		  ]
+		}`,
+			ExpectedPolicy: schedulerapi.Policy{
+				Predicates: []schedulerapi.PredicatePolicy{
+					{Name: "MatchNodeSelector"},
+					{Name: "PodFitsResources"},
+					{Name: "PodFitsHostPorts"},
+					{Name: "HostName"},
+					{Name: "NoDiskConflict"},
+					{Name: "NoVolumeZoneConflict"},
+					{Name: "PodToleratesNodeTaints"},
+					{Name: "CheckNodeMemoryPressure"},
+					{Name: "CheckNodeDiskPressure"},
+					{Name: "CheckNodeCondition"},
+					{Name: "MaxEBSVolumeCount"},
+					{Name: "MaxGCEPDVolumeCount"},
+					{Name: "MaxAzureDiskVolumeCount"},
+					{Name: "MatchInterPodAffinity"},
+					{Name: "GeneralPredicates"},
+					{Name: "TestServiceAffinity", Argument: &schedulerapi.PredicateArgument{ServiceAffinity: &schedulerapi.ServiceAffinity{Labels: []string{"region"}}}},
+					{Name: "TestLabelsPresence", Argument: &schedulerapi.PredicateArgument{LabelsPresence: &schedulerapi.LabelsPresence{Labels: []string{"foo"}, Presence: true}}},
+					{Name: "NoVolumeNodeConflict"},
+				},
+				Priorities: []schedulerapi.PriorityPolicy{
+					{Name: "EqualPriority", Weight: 2},
+					{Name: "ImageLocalityPriority", Weight: 2},
+					{Name: "LeastRequestedPriority", Weight: 2},
+					{Name: "BalancedResourceAllocation", Weight: 2},
+					{Name: "SelectorSpreadPriority", Weight: 2},
+					{Name: "NodePreferAvoidPodsPriority", Weight: 2},
+					{Name: "NodeAffinityPriority", Weight: 2},
+					{Name: "TaintTolerationPriority", Weight: 2},
+					{Name: "InterPodAffinityPriority", Weight: 2},
+					{Name: "MostRequestedPriority", Weight: 2},
+				},
+			},
+		},
 	}

 	registeredPredicates := sets.NewString(factory.ListRegisteredFitPredicates()...)
--- a/plugin/pkg/scheduler/algorithmprovider/defaults/defaults.go
+++ b/plugin/pkg/scheduler/algorithmprovider/defaults/defaults.go
@@ -175,6 +175,9 @@ func defaultPredicates() sets.String {
 		// Fit is determined by node disk pressure condition.
 		factory.RegisterFitPredicate("CheckNodeDiskPressure", predicates.CheckNodeDiskPressurePredicate),

+		// Fit is determied by node condtions: not ready, network unavailable and out of disk.
+		factory.RegisterMandatoryFitPredicate("CheckNodeCondition", predicates.CheckNodeConditionPredicate),
+
 		// Fit is determined by volume zone requirements.
 		factory.RegisterFitPredicateFactory(
 			"NoVolumeNodeConflict",
--- a/plugin/pkg/scheduler/factory/factory.go
+++ b/plugin/pkg/scheduler/factory/factory.go
@@ -705,7 +705,7 @@ func (f *ConfigFactory) CreateFromKeys(predicateKeys, priorityKeys sets.String,
 		SchedulerCache: f.schedulerCache,
 		Ecache:         f.equivalencePodCache,
 		// The scheduler only needs to consider schedulable nodes.
-		NodeLister:          &nodePredicateLister{f.nodeLister},
+		NodeLister:          &nodeLister{f.nodeLister},
 		Algorithm:           algo,
 		Binder:              f.getBinder(extenders),
 		PodConditionUpdater: &podConditionUpdater{f.client},
@@ -720,12 +720,12 @@ func (f *ConfigFactory) CreateFromKeys(predicateKeys, priorityKeys sets.String,
 	}, nil
 }

-type nodePredicateLister struct {
+type nodeLister struct {
 	corelisters.NodeLister
 }

-func (n *nodePredicateLister) List() ([]*v1.Node, error) {
-	return n.ListWithPredicate(getNodeConditionPredicate())
+func (n *nodeLister) List() ([]*v1.Node, error) {
+	return n.NodeLister.List(labels.Everything())
 }

 func (f *ConfigFactory) GetPriorityFunctionConfigs(priorityKeys sets.String) ([]algorithm.PriorityConfig, error) {
@@ -770,8 +770,7 @@ func (f *ConfigFactory) getPluginArgs() (*PluginFactoryArgs, error) {
 		ControllerLister:  f.controllerLister,
 		ReplicaSetLister:  f.replicaSetLister,
 		StatefulSetLister: f.statefulSetLister,
-		// All fit predicates only need to consider schedulable nodes.
-		NodeLister: &nodePredicateLister{f.nodeLister},
+		NodeLister:        &nodeLister{f.nodeLister},
 		NodeInfo:          &predicates.CachedNodeInfo{NodeLister: f.nodeLister},
 		PVInfo:            &predicates.CachedPersistentVolumeInfo{PersistentVolumeLister: f.pVLister},
 		PVCInfo:           &predicates.CachedPersistentVolumeClaimInfo{PersistentVolumeClaimLister: f.pVCLister},
@@ -793,34 +792,6 @@ func (f *ConfigFactory) ResponsibleForPod(pod *v1.Pod) bool {
 	return f.schedulerName == pod.Spec.SchedulerName
 }

-func getNodeConditionPredicate() corelisters.NodeConditionPredicate {
-	return func(node *v1.Node) bool {
-		for i := range node.Status.Conditions {
-			cond := &node.Status.Conditions[i]
-			// We consider the node for scheduling only when its:
-			// - NodeReady condition status is ConditionTrue,
-			// - NodeOutOfDisk condition status is ConditionFalse,
-			// - NodeNetworkUnavailable condition status is ConditionFalse.
-			if cond.Type == v1.NodeReady && cond.Status != v1.ConditionTrue {
-				glog.V(4).Infof("Ignoring node %v with %v condition status %v", node.Name, cond.Type, cond.Status)
-				return false
-			} else if cond.Type == v1.NodeOutOfDisk && cond.Status != v1.ConditionFalse {
-				glog.V(4).Infof("Ignoring node %v with %v condition status %v", node.Name, cond.Type, cond.Status)
-				return false
-			} else if cond.Type == v1.NodeNetworkUnavailable && cond.Status != v1.ConditionFalse {
-				glog.V(4).Infof("Ignoring node %v with %v condition status %v", node.Name, cond.Type, cond.Status)
-				return false
-			}
-		}
-		// Ignore nodes that are marked unschedulable
-		if node.Spec.Unschedulable {
-			glog.V(4).Infof("Ignoring node %v since it is unschedulable", node.Name)
-			return false
-		}
-		return true
-	}
-}
-
 // unassignedNonTerminatedPod selects pods that are unassigned and non-terminal.
 func unassignedNonTerminatedPod(pod *v1.Pod) bool {
 	if len(pod.Spec.NodeName) != 0 {
--- a/plugin/pkg/scheduler/factory/factory_test.go
+++ b/plugin/pkg/scheduler/factory/factory_test.go
@@ -525,46 +525,3 @@ func TestInvalidFactoryArgs(t *testing.T) {
 	}

 }
-
-func TestNodeConditionPredicate(t *testing.T) {
-	nodeFunc := getNodeConditionPredicate()
-	nodeList := &v1.NodeList{
-		Items: []v1.Node{
-			// node1 considered
-			{ObjectMeta: metav1.ObjectMeta{Name: "node1"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionTrue}}}},
-			// node2 ignored - node not Ready
-			{ObjectMeta: metav1.ObjectMeta{Name: "node2"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionFalse}}}},
-			// node3 ignored - node out of disk
-			{ObjectMeta: metav1.ObjectMeta{Name: "node3"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeOutOfDisk, Status: v1.ConditionTrue}}}},
-			// node4 considered
-			{ObjectMeta: metav1.ObjectMeta{Name: "node4"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeOutOfDisk, Status: v1.ConditionFalse}}}},
-
-			// node5 ignored - node out of disk
-			{ObjectMeta: metav1.ObjectMeta{Name: "node5"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionTrue}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionTrue}}}},
-			// node6 considered
-			{ObjectMeta: metav1.ObjectMeta{Name: "node6"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionTrue}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionFalse}}}},
-			// node7 ignored - node out of disk, node not Ready
-			{ObjectMeta: metav1.ObjectMeta{Name: "node7"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionFalse}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionTrue}}}},
-			// node8 ignored - node not Ready
-			{ObjectMeta: metav1.ObjectMeta{Name: "node8"}, Status: v1.NodeStatus{Conditions: []v1.NodeCondition{{Type: v1.NodeReady, Status: v1.ConditionFalse}, {Type: v1.NodeOutOfDisk, Status: v1.ConditionFalse}}}},
-
-			// node9 ignored - node unschedulable
-			{ObjectMeta: metav1.ObjectMeta{Name: "node9"}, Spec: v1.NodeSpec{Unschedulable: true}},
-			// node10 considered
-			{ObjectMeta: metav1.ObjectMeta{Name: "node10"}, Spec: v1.NodeSpec{Unschedulable: false}},
-			// node11 considered
-			{ObjectMeta: metav1.ObjectMeta{Name: "node11"}},
-		},
-	}
-
-	nodeNames := []string{}
-	for _, node := range nodeList.Items {
-		if nodeFunc(&node) {
-			nodeNames = append(nodeNames, node.Name)
-		}
-	}
-	expectedNodes := []string{"node1", "node4", "node6", "node10", "node11"}
-	if !reflect.DeepEqual(expectedNodes, nodeNames) {
-		t.Errorf("expected: %v, got %v", expectedNodes, nodeNames)
-	}
-}
--- a/plugin/pkg/scheduler/factory/plugins.go
+++ b/plugin/pkg/scheduler/factory/plugins.go
@@ -73,6 +73,7 @@ var (

 	// maps that hold registered algorithm types
 	fitPredicateMap        = make(map[string]FitPredicateFactory)
+	mandatoryFitPredicates = sets.NewString()
 	priorityFunctionMap    = make(map[string]PriorityConfigFactory)
 	algorithmProviderMap   = make(map[string]AlgorithmProviderConfig)

@@ -99,6 +100,18 @@ func RegisterFitPredicate(name string, predicate algorithm.FitPredicate) string
 	return RegisterFitPredicateFactory(name, func(PluginFactoryArgs) algorithm.FitPredicate { return predicate })
 }

+// RegisterMandatoryFitPredicate registers a fit predicate with the algorithm registry, the predicate is used by
+// kubelet, DaemonSet; it is always included in configuration. Returns the name with which the predicate was
+// registered.
+func RegisterMandatoryFitPredicate(name string, predicate algorithm.FitPredicate) string {
+	schedulerFactoryMutex.Lock()
+	defer schedulerFactoryMutex.Unlock()
+	validateAlgorithmNameOrDie(name)
+	fitPredicateMap[name] = func(PluginFactoryArgs) algorithm.FitPredicate { return predicate }
+	mandatoryFitPredicates.Insert(name)
+	return name
+}
+
 // RegisterFitPredicateFactory registers a fit predicate factory with the
 // algorithm registry. Returns the name with which the predicate was registered.
 func RegisterFitPredicateFactory(name string, predicateFactory FitPredicateFactory) string {
@@ -308,6 +321,14 @@ func getFitPredicateFunctions(names sets.String, args PluginFactoryArgs) (map[st
 		}
 		predicates[name] = factory(args)
 	}
+
+	// Always include mandatory fit predicates.
+	for name := range mandatoryFitPredicates {
+		if factory, found := fitPredicateMap[name]; found {
+			predicates[name] = factory(args)
+		}
+	}
+
 	return predicates, nil
 }

--- a/test/integration/scheduler/scheduler_test.go
+++ b/test/integration/scheduler/scheduler_test.go
@@ -139,7 +139,8 @@ func TestSchedulerCreationFromConfigMap(t *testing.T) {
 	// Verify that the config is applied correctly.
 	schedPredicates := sched.Config().Algorithm.Predicates()
 	schedPrioritizers := sched.Config().Algorithm.Prioritizers()
-	if len(schedPredicates) != 2 || len(schedPrioritizers) != 2 {
+	// Includes one mandatory predicates.
+	if len(schedPredicates) != 3 || len(schedPrioritizers) != 2 {
 		t.Errorf("Unexpected number of predicates or priority functions. Number of predicates: %v, number of prioritizers: %v", len(schedPredicates), len(schedPrioritizers))
 	}
 	// Check a predicate and a priority function.