Use PodTopologySpread for default spreading

Under the feature gate DefaultPodTopologySpread, which will disable the legacy DefaultPodTopologySpread plugin from the default algorithm providers.

Signed-off-by: Aldo Culquicondor <acondor@google.com>

pkg/scheduler/algorithmprovider/BUILD

@@ -11,6 +11,7 @@ go_library(
     srcs = ["registry.go"],
     importpath = "k8s.io/kubernetes/pkg/scheduler/algorithmprovider",
     deps = [
+        "//pkg/features:go_default_library",
         "//pkg/scheduler/apis/config:go_default_library",
         "//pkg/scheduler/framework/plugins/defaultbinder:go_default_library",
         "//pkg/scheduler/framework/plugins/defaultpodtopologyspread:go_default_library",
@@ -29,6 +30,8 @@ go_library(
         "//pkg/scheduler/framework/plugins/volumebinding:go_default_library",
         "//pkg/scheduler/framework/plugins/volumerestrictions:go_default_library",
         "//pkg/scheduler/framework/plugins/volumezone:go_default_library",
+        "//staging/src/k8s.io/apiserver/pkg/util/feature:go_default_library",
+        "//vendor/k8s.io/klog/v2:go_default_library",
     ],
 )
 
@@ -37,6 +40,7 @@ go_test(
     srcs = ["registry_test.go"],
     embed = [":go_default_library"],
     deps = [
+        "//pkg/features:go_default_library",
         "//pkg/scheduler/apis/config:go_default_library",
         "//pkg/scheduler/framework/plugins/defaultbinder:go_default_library",
         "//pkg/scheduler/framework/plugins/defaultpodtopologyspread:go_default_library",
@@ -55,6 +59,9 @@ go_test(
         "//pkg/scheduler/framework/plugins/volumebinding:go_default_library",
         "//pkg/scheduler/framework/plugins/volumerestrictions:go_default_library",
         "//pkg/scheduler/framework/plugins/volumezone:go_default_library",
+        "//staging/src/k8s.io/apiserver/pkg/util/feature:go_default_library",
+        "//staging/src/k8s.io/component-base/featuregate:go_default_library",
+        "//staging/src/k8s.io/component-base/featuregate/testing:go_default_library",
         "//vendor/github.com/google/go-cmp/cmp:go_default_library",
     ],
 )

pkg/scheduler/algorithmprovider/registry.go

@@ -20,6 +20,9 @@ import (
 	"sort"
 	"strings"
 
+	utilfeature "k8s.io/apiserver/pkg/util/feature"
+	"k8s.io/klog/v2"
+	"k8s.io/kubernetes/pkg/features"
 	schedulerapi "k8s.io/kubernetes/pkg/scheduler/apis/config"
 	"k8s.io/kubernetes/pkg/scheduler/framework/plugins/defaultbinder"
 	"k8s.io/kubernetes/pkg/scheduler/framework/plugins/defaultpodtopologyspread"
@@ -49,7 +52,10 @@ type Registry map[string]*schedulerapi.Plugins
 // NewRegistry returns an algorithm provider registry instance.
 func NewRegistry() Registry {
 	defaultConfig := getDefaultConfig()
+	applyFeatureGates(defaultConfig)
+
 	caConfig := getClusterAutoscalerConfig()
+	applyFeatureGates(caConfig)
 
 	return Registry{
 		schedulerapi.SchedulerDefaultProviderName: defaultConfig,
@@ -106,7 +112,6 @@ func getDefaultConfig() *schedulerapi.Plugins {
 			Enabled: []schedulerapi.Plugin{
 				{Name: interpodaffinity.Name},
 				{Name: podtopologyspread.Name},
-				{Name: defaultpodtopologyspread.Name},
 				{Name: tainttoleration.Name},
 			},
 		},
@@ -122,7 +127,6 @@ func getDefaultConfig() *schedulerapi.Plugins {
 				// - This is a score coming from user preference.
 				// - It makes its signal comparable to NodeResourcesLeastAllocated.
 				{Name: podtopologyspread.Name, Weight: 2},
-				{Name: defaultpodtopologyspread.Name, Weight: 1},
 				{Name: tainttoleration.Name, Weight: 1},
 			},
 		},
@@ -164,3 +168,15 @@ func getClusterAutoscalerConfig() *schedulerapi.Plugins {
 	}
 	return caConfig
 }
+
+func applyFeatureGates(config *schedulerapi.Plugins) {
+	if !utilfeature.DefaultFeatureGate.Enabled(features.DefaultPodTopologySpread) {
+		// When feature is enabled, the default spreading is done by
+		// PodTopologySpread plugin, which is enabled by default.
+		klog.Infof("Registering DefaultPodTopologySpread plugin")
+		s := schedulerapi.Plugin{Name: defaultpodtopologyspread.Name}
+		config.PreScore.Enabled = append(config.PreScore.Enabled, s)
+		s.Weight = 1
+		config.Score.Enabled = append(config.Score.Enabled, s)
+	}
+}

pkg/scheduler/algorithmprovider/registry_test.go

@@ -20,7 +20,11 @@ import (
 	"testing"
 
 	"github.com/google/go-cmp/cmp"
+	"k8s.io/component-base/featuregate"
 
+	utilfeature "k8s.io/apiserver/pkg/util/feature"
+	featuregatetesting "k8s.io/component-base/featuregate/testing"
+	"k8s.io/kubernetes/pkg/features"
 	schedulerapi "k8s.io/kubernetes/pkg/scheduler/apis/config"
 	"k8s.io/kubernetes/pkg/scheduler/framework/plugins/defaultbinder"
 	"k8s.io/kubernetes/pkg/scheduler/framework/plugins/defaultpodtopologyspread"
@@ -79,8 +83,8 @@ func TestClusterAutoscalerProvider(t *testing.T) {
 			Enabled: []schedulerapi.Plugin{
 				{Name: interpodaffinity.Name},
 				{Name: podtopologyspread.Name},
-				{Name: defaultpodtopologyspread.Name},
 				{Name: tainttoleration.Name},
+				{Name: defaultpodtopologyspread.Name},
 			},
 		},
 		Score: &schedulerapi.PluginSet{
@@ -92,8 +96,8 @@ func TestClusterAutoscalerProvider(t *testing.T) {
 				{Name: nodeaffinity.Name, Weight: 1},
 				{Name: nodepreferavoidpods.Name, Weight: 10000},
 				{Name: podtopologyspread.Name, Weight: 2},
-				{Name: defaultpodtopologyspread.Name, Weight: 1},
 				{Name: tainttoleration.Name, Weight: 1},
+				{Name: defaultpodtopologyspread.Name, Weight: 1},
 			},
 		},
 		Reserve: &schedulerapi.PluginSet{
@@ -129,3 +133,191 @@ func TestClusterAutoscalerProvider(t *testing.T) {
 		t.Errorf("unexpected config diff (-want, +got): %s", diff)
 	}
 }
+
+func TestApplyFeatureGates(t *testing.T) {
+	tests := []struct {
+		name       string
+		feature    featuregate.Feature
+		wantConfig *schedulerapi.Plugins
+	}{
+		{
+			name: "Feature gates disabled",
+			wantConfig: &schedulerapi.Plugins{
+				QueueSort: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: queuesort.Name},
+					},
+				},
+				PreFilter: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: noderesources.FitName},
+						{Name: nodeports.Name},
+						{Name: podtopologyspread.Name},
+						{Name: interpodaffinity.Name},
+					},
+				},
+				Filter: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: nodeunschedulable.Name},
+						{Name: noderesources.FitName},
+						{Name: nodename.Name},
+						{Name: nodeports.Name},
+						{Name: nodeaffinity.Name},
+						{Name: volumerestrictions.Name},
+						{Name: tainttoleration.Name},
+						{Name: nodevolumelimits.EBSName},
+						{Name: nodevolumelimits.GCEPDName},
+						{Name: nodevolumelimits.CSIName},
+						{Name: nodevolumelimits.AzureDiskName},
+						{Name: volumebinding.Name},
+						{Name: volumezone.Name},
+						{Name: podtopologyspread.Name},
+						{Name: interpodaffinity.Name},
+					},
+				},
+				PreScore: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: interpodaffinity.Name},
+						{Name: podtopologyspread.Name},
+						{Name: tainttoleration.Name},
+						{Name: defaultpodtopologyspread.Name},
+					},
+				},
+				Score: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: noderesources.BalancedAllocationName, Weight: 1},
+						{Name: imagelocality.Name, Weight: 1},
+						{Name: interpodaffinity.Name, Weight: 1},
+						{Name: noderesources.LeastAllocatedName, Weight: 1},
+						{Name: nodeaffinity.Name, Weight: 1},
+						{Name: nodepreferavoidpods.Name, Weight: 10000},
+						{Name: podtopologyspread.Name, Weight: 2},
+						{Name: tainttoleration.Name, Weight: 1},
+						{Name: defaultpodtopologyspread.Name, Weight: 1},
+					},
+				},
+				Reserve: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+				Unreserve: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+				PreBind: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+				Bind: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: defaultbinder.Name},
+					},
+				},
+				PostBind: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+			},
+		},
+		{
+			name:    "NewDefaultPodTopologySpread enabled",
+			feature: features.DefaultPodTopologySpread,
+			wantConfig: &schedulerapi.Plugins{
+				QueueSort: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: queuesort.Name},
+					},
+				},
+				PreFilter: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: noderesources.FitName},
+						{Name: nodeports.Name},
+						{Name: podtopologyspread.Name},
+						{Name: interpodaffinity.Name},
+					},
+				},
+				Filter: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: nodeunschedulable.Name},
+						{Name: noderesources.FitName},
+						{Name: nodename.Name},
+						{Name: nodeports.Name},
+						{Name: nodeaffinity.Name},
+						{Name: volumerestrictions.Name},
+						{Name: tainttoleration.Name},
+						{Name: nodevolumelimits.EBSName},
+						{Name: nodevolumelimits.GCEPDName},
+						{Name: nodevolumelimits.CSIName},
+						{Name: nodevolumelimits.AzureDiskName},
+						{Name: volumebinding.Name},
+						{Name: volumezone.Name},
+						{Name: podtopologyspread.Name},
+						{Name: interpodaffinity.Name},
+					},
+				},
+				PreScore: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: interpodaffinity.Name},
+						{Name: podtopologyspread.Name},
+						{Name: tainttoleration.Name},
+					},
+				},
+				Score: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: noderesources.BalancedAllocationName, Weight: 1},
+						{Name: imagelocality.Name, Weight: 1},
+						{Name: interpodaffinity.Name, Weight: 1},
+						{Name: noderesources.LeastAllocatedName, Weight: 1},
+						{Name: nodeaffinity.Name, Weight: 1},
+						{Name: nodepreferavoidpods.Name, Weight: 10000},
+						{Name: podtopologyspread.Name, Weight: 2},
+						{Name: tainttoleration.Name, Weight: 1},
+					},
+				},
+				Reserve: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+				Unreserve: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+				PreBind: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+				Bind: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: defaultbinder.Name},
+					},
+				},
+				PostBind: &schedulerapi.PluginSet{
+					Enabled: []schedulerapi.Plugin{
+						{Name: volumebinding.Name},
+					},
+				},
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			if test.feature != "" {
+				defer featuregatetesting.SetFeatureGateDuringTest(t, utilfeature.DefaultFeatureGate, test.feature, true)()
+			}
+
+			r := NewRegistry()
+			gotConfig := r[schedulerapi.SchedulerDefaultProviderName]
+			if diff := cmp.Diff(test.wantConfig, gotConfig); diff != "" {
+				t.Errorf("unexpected config diff (-want, +got): %s", diff)
+			}
+		})
+	}
+}