Merge pull request #1427 from a7i/helm-v0.30.1-release

helm: upgrade to v0.30.1

README.md

@@ -2,7 +2,7 @@
 ![Release Charts](https://github.com/kubernetes-sigs/descheduler/workflows/Release%20Charts/badge.svg)
 
 <p align="left">
-	↖️ Click at the [bullet list icon] at the top left corner of the Readme visualization for the github generated table of contents.
+	↗️️ Click at the [bullet list icon] at the top right corner of the Readme visualization for the github generated table of contents.
 </p>
 
 <p align="center">

charts/descheduler/Chart.yaml

@@ -1,7 +1,7 @@
 apiVersion: v1
 name: descheduler
-version: 0.30.0
+version: 0.30.1
-appVersion: 0.30.0
+appVersion: 0.30.1
 description: Descheduler for Kubernetes is used to rebalance clusters by evicting pods that can potentially be scheduled on better nodes. In the current implementation, descheduler does not schedule replacement of evicted pods but relies on the default scheduler for that.
 keywords:
 - kubernetes

charts/descheduler/templates/NOTES.txt

@@ -1,7 +1,7 @@
 Descheduler installed as a {{ .Values.kind }}.
 
 {{- if eq .Values.kind "Deployment" }}
-{{- if eq .Values.replicas 1.0}}
+{{- if eq (.Values.replicas | int) 1 }}
 WARNING: You set replica count as 1 and workload kind as Deployment however leaderElection is not enabled. Consider enabling Leader Election for HA mode.
 {{- end}}
 {{- if .Values.leaderElection }}

charts/descheduler/templates/cronjob.yaml

@@ -81,7 +81,11 @@ spec:
               args:
                 - --policy-config-file=/policy-dir/policy.yaml
                 {{- range $key, $value := .Values.cmdOptions }}
-                - {{ printf "--%s" $key }}{{ if $value }}={{ $value }}{{ end }}
+                {{- if ne $value nil }}
+                - {{ printf "--%s=%s" $key (toString $value) }}
+                {{- else }}
+                - {{ printf "--%s" $key }}
+                {{- end }}
                 {{- end }}
               livenessProbe:
                 {{- toYaml .Values.livenessProbe | nindent 16 }}

charts/descheduler/templates/deployment.yaml

@@ -7,7 +7,7 @@ metadata:
   labels:
     {{- include "descheduler.labels" . | nindent 4 }}
 spec:
-  {{- if gt .Values.replicas 1.0}}
+  {{- if gt (.Values.replicas | int) 1 }}
   {{- if not .Values.leaderElection.enabled }}
   {{- fail "You must set leaderElection to use more than 1 replica"}}
   {{- end}}
@@ -53,7 +53,11 @@ spec:
             - --policy-config-file=/policy-dir/policy.yaml
             - --descheduling-interval={{ required "deschedulingInterval required for running as Deployment" .Values.deschedulingInterval }}
             {{- range $key, $value := .Values.cmdOptions }}
-            - {{ printf "--%s" $key }}{{ if $value }}={{ $value }}{{ end }}
+            {{- if ne $value nil }}
+            - {{ printf "--%s=%s" $key (toString $value) }}
+            {{- else }}
+            - {{ printf "--%s" $key }}
+            {{- end }}
             {{- end }}
             {{- include "descheduler.leaderElection" . | nindent 12 }}
           ports:

charts/descheduler/values.yaml

@@ -111,14 +111,13 @@ deschedulerPolicy:
           args:
             podRestartThreshold: 100
             includingInitContainers: true
-        - name: RemovePodsViolatingNodeTaints
+        - name: RemovePodsViolatingNodeAffinity
           args:
             nodeAffinityType:
-              - requiredDuringSchedulingIgnoredDuringExecution
+            - requiredDuringSchedulingIgnoredDuringExecution
+        - name: RemovePodsViolatingNodeTaints
         - name: RemovePodsViolatingInterPodAntiAffinity
         - name: RemovePodsViolatingTopologySpreadConstraint
-          args:
-            includeSoftConstraints: false
         - name: LowNodeUtilization
           args:
             thresholds:
@@ -133,7 +132,6 @@ deschedulerPolicy:
         balance:
           enabled:
             - RemoveDuplicates
-            - RemovePodsViolatingNodeAffinity
             - RemovePodsViolatingTopologySpreadConstraint
             - LowNodeUtilization
         deschedule:

cmd/descheduler/app/server.go

@@ -77,7 +77,7 @@ func NewDeschedulerCommand(out io.Writer) *cobra.Command {
 
 			secureServing.DisableHTTP2 = !s.EnableHTTP2
 
-			ctx, done := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
+			ctx, done := signal.NotifyContext(cmd.Context(), syscall.SIGINT, syscall.SIGTERM)
 
 			pathRecorderMux := mux.NewPathRecorderMux("descheduler")
 			if !s.DisableMetrics {

docs/deprecated/v1alpha1.md

@@ -109,17 +109,17 @@ See the [resources | Kustomize](https://kubectl.docs.kubernetes.io/references/ku
 
 Run As A Job
 ```
-kustomize build 'github.com/kubernetes-sigs/descheduler/kubernetes/job?ref=v0.30.0' | kubectl apply -f -
+kustomize build 'github.com/kubernetes-sigs/descheduler/kubernetes/job?ref=v0.30.1' | kubectl apply -f -
 ```
 
 Run As A CronJob
 ```
-kustomize build 'github.com/kubernetes-sigs/descheduler/kubernetes/cronjob?ref=v0.30.0' | kubectl apply -f -
+kustomize build 'github.com/kubernetes-sigs/descheduler/kubernetes/cronjob?ref=v0.30.1' | kubectl apply -f -
 ```
 
 Run As A Deployment
 ```
-kustomize build 'github.com/kubernetes-sigs/descheduler/kubernetes/deployment?ref=v0.30.0' | kubectl apply -f -
+kustomize build 'github.com/kubernetes-sigs/descheduler/kubernetes/deployment?ref=v0.30.1' | kubectl apply -f -
 ```
 
 ## User Guide

docs/user-guide.md

@@ -4,6 +4,7 @@ Starting with descheduler release v0.10.0 container images are available in the
 
 Descheduler Version | Container Image                                 | Architectures           |
 ------------------- |-------------------------------------------------|-------------------------|
+v0.30.1             | registry.k8s.io/descheduler/descheduler:v0.30.1 | AMD64<br>ARM64<br>ARMv7 |
 v0.30.0             | registry.k8s.io/descheduler/descheduler:v0.30.0 | AMD64<br>ARM64<br>ARMv7 |
 v0.29.0             | registry.k8s.io/descheduler/descheduler:v0.29.0 | AMD64<br>ARM64<br>ARMv7 |
 v0.28.1             | registry.k8s.io/descheduler/descheduler:v0.28.1 | AMD64<br>ARM64<br>ARMv7 |

kubernetes/cronjob/cronjob.yaml

@@ -16,7 +16,7 @@ spec:
           priorityClassName: system-cluster-critical
           containers:
           - name: descheduler
-            image: registry.k8s.io/descheduler/descheduler:v0.30.0
+            image: registry.k8s.io/descheduler/descheduler:v0.30.1
             volumeMounts:
             - mountPath: /policy-dir
               name: policy-volume

kubernetes/deployment/deployment.yaml

@@ -19,7 +19,7 @@ spec:
       serviceAccountName: descheduler-sa
       containers:
         - name: descheduler
-          image: registry.k8s.io/descheduler/descheduler:v0.30.0
+          image: registry.k8s.io/descheduler/descheduler:v0.30.1
           imagePullPolicy: IfNotPresent
           command:
             - "/bin/descheduler"

kubernetes/job/job.yaml

@@ -14,7 +14,7 @@ spec:
       priorityClassName: system-cluster-critical
       containers:
         - name: descheduler
-          image: registry.k8s.io/descheduler/descheduler:v0.30.0
+          image: registry.k8s.io/descheduler/descheduler:v0.30.1
           volumeMounts:
           - mountPath: /policy-dir
             name: policy-volume

pkg/descheduler/node/node.go

@@ -343,9 +343,30 @@ func podMatchesInterPodAntiAffinity(nodeIndexer podutil.GetPodsAssignedToNodeFun
 	if err != nil {
 		return false, fmt.Errorf("error listing all pods: %v", err)
 	}
+	assignedPodsInNamespace := podutil.GroupByNamespace(podsOnNode)
 
-	podsInANamespace := podutil.GroupByNamespace(podsOnNode)
+	for _, term := range utils.GetPodAntiAffinityTerms(pod.Spec.Affinity.PodAntiAffinity) {
-	nodeMap := utils.CreateNodeMap([]*v1.Node{node})
+		namespaces := utils.GetNamespacesFromPodAffinityTerm(pod, &term)
+		selector, err := metav1.LabelSelectorAsSelector(term.LabelSelector)
+		if err != nil {
+			klog.ErrorS(err, "Unable to convert LabelSelector into Selector")
+			return false, err
+		}
 
-	return utils.CheckPodsWithAntiAffinityExist(pod, podsInANamespace, nodeMap), nil
+		for namespace := range namespaces {
+			for _, assignedPod := range assignedPodsInNamespace[namespace] {
+				if assignedPod.Name == pod.Name || !utils.PodMatchesTermsNamespaceAndSelector(assignedPod, namespaces, selector) {
+					klog.V(4).InfoS("Pod doesn't match inter-pod anti-affinity rule of assigned pod on node", "candidatePod", klog.KObj(pod), "assignedPod", klog.KObj(assignedPod))
+					continue
+				}
+
+				if _, ok := node.Labels[term.TopologyKey]; ok {
+					klog.V(1).InfoS("Pod matches inter-pod anti-affinity rule of assigned pod on node", "candidatePod", klog.KObj(pod), "assignedPod", klog.KObj(assignedPod))
+					return true, nil
+				}
+			}
+		}
+	}
+
+	return false, nil
 }

pkg/descheduler/node/node_test.go

@@ -759,6 +759,9 @@ func TestNodeFit(t *testing.T) {
 			"region": "main-region",
 		}
 	})
+
+	nodeNolabel := test.BuildTestNode("node", 64000, 128*1000*1000*1000, 2, nil)
+
 	tests := []struct {
 		description string
 		pod         *v1.Pod
@@ -767,7 +770,7 @@ func TestNodeFit(t *testing.T) {
 		err         error
 	}{
 		{
-			description: "insufficient cpu",
+			description: "Insufficient cpu",
 			pod:         test.BuildTestPod("p1", 10000, 2*1000*1000*1000, "", nil),
 			node:        node,
 			podsOnNode: []*v1.Pod{
@@ -776,7 +779,7 @@ func TestNodeFit(t *testing.T) {
 			err: errors.New("insufficient cpu"),
 		},
 		{
-			description: "insufficient pod num",
+			description: "Insufficient pod num",
 			pod:         test.BuildTestPod("p1", 1000, 2*1000*1000*1000, "", nil),
 			node:        node,
 			podsOnNode: []*v1.Pod{
@@ -786,7 +789,7 @@ func TestNodeFit(t *testing.T) {
 			err: errors.New("insufficient pods"),
 		},
 		{
-			description: "matches inter-pod anti-affinity rule of pod on node",
+			description: "Pod matches inter-pod anti-affinity rule of other pod on node",
 			pod:         test.PodWithPodAntiAffinity(test.BuildTestPod("p1", 1000, 1000, node.Name, nil), "foo", "bar"),
 			node:        node,
 			podsOnNode: []*v1.Pod{
@@ -795,11 +798,36 @@ func TestNodeFit(t *testing.T) {
 			err: errors.New("pod matches inter-pod anti-affinity rule of other pod on node"),
 		},
 		{
-			description: "pod fits on node",
+			description: "Pod doesn't match inter-pod anti-affinity rule of other pod on node, because pod and other pod is not same namespace",
+			pod:         test.PodWithPodAntiAffinity(test.BuildTestPod("p1", 1000, 1000, node.Name, nil), "foo", "bar"),
+			node:        node,
+			podsOnNode: []*v1.Pod{
+				test.PodWithPodAntiAffinity(test.BuildTestPod("p2", 1000, 1000, node.Name, func(pod *v1.Pod) {
+					pod.Namespace = "test"
+				}), "foo", "bar"),
+			},
+		},
+		{
+			description: "Pod doesn't match inter-pod anti-affinity rule of other pod on node, because other pod not match labels of pod",
+			pod:         test.PodWithPodAntiAffinity(test.BuildTestPod("p1", 1000, 1000, node.Name, nil), "foo", "bar"),
+			node:        node,
+			podsOnNode: []*v1.Pod{
+				test.PodWithPodAntiAffinity(test.BuildTestPod("p2", 1000, 1000, node.Name, nil), "foo1", "bar1"),
+			},
+		},
+		{
+			description: "Pod doesn't match inter-pod anti-affinity rule of other pod on node, because node have no topologyKey",
+			pod:         test.PodWithPodAntiAffinity(test.BuildTestPod("p1", 1000, 1000, "node1", nil), "foo", "bar"),
+			node:        nodeNolabel,
+			podsOnNode: []*v1.Pod{
+				test.PodWithPodAntiAffinity(test.BuildTestPod("p2", 1000, 1000, node.Name, nil), "foo", "bar"),
+			},
+		},
+		{
+			description: "Pod fits on node",
 			pod:         test.BuildTestPod("p1", 1000, 1000, "", func(pod *v1.Pod) {}),
 			node:        node,
 			podsOnNode:  []*v1.Pod{},
-			err:         nil,
 		},
 	}
 

pkg/utils/predicates.go

@@ -24,10 +24,37 @@ import (
 	v1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/labels"
+	"k8s.io/apimachinery/pkg/util/sets"
 	"k8s.io/component-helpers/scheduling/corev1"
 	"k8s.io/klog/v2"
 )
 
+// GetNamespacesFromPodAffinityTerm returns a set of names
+// according to the namespaces indicated in podAffinityTerm.
+// If namespaces is empty it considers the given pod's namespace.
+func GetNamespacesFromPodAffinityTerm(pod *v1.Pod, podAffinityTerm *v1.PodAffinityTerm) sets.Set[string] {
+	names := sets.New[string]()
+	if len(podAffinityTerm.Namespaces) == 0 {
+		names.Insert(pod.Namespace)
+	} else {
+		names.Insert(podAffinityTerm.Namespaces...)
+	}
+	return names
+}
+
+// PodMatchesTermsNamespaceAndSelector returns true if the given <pod>
+// matches the namespace and selector defined by <affinityPod>`s <term>.
+func PodMatchesTermsNamespaceAndSelector(pod *v1.Pod, namespaces sets.Set[string], selector labels.Selector) bool {
+	if !namespaces.Has(pod.Namespace) {
+		return false
+	}
+
+	if !selector.Matches(labels.Set(pod.Labels)) {
+		return false
+	}
+	return true
+}
+
 // The following code has been copied from predicates package to avoid the
 // huge vendoring issues, mostly copied from
 // k8s.io/kubernetes/plugin/pkg/scheduler/algorithm/predicates/
@@ -309,42 +336,52 @@ func CreateNodeMap(nodes []*v1.Node) map[string]*v1.Node {
 	return m
 }
 
-// CheckPodsWithAntiAffinityExist checks if there are other pods on the node that the current pod cannot tolerate.
+// CheckPodsWithAntiAffinityExist checks if there are other pods on the node that the current candidate pod cannot tolerate.
-func CheckPodsWithAntiAffinityExist(pod *v1.Pod, pods map[string][]*v1.Pod, nodeMap map[string]*v1.Node) bool {
+func CheckPodsWithAntiAffinityExist(candidatePod *v1.Pod, assignedPods map[string][]*v1.Pod, nodeMap map[string]*v1.Node) bool {
-	affinity := pod.Spec.Affinity
+	nodeHavingCandidatePod, ok := nodeMap[candidatePod.Spec.NodeName]
-	if affinity != nil && affinity.PodAntiAffinity != nil {
+	if !ok {
-		for _, term := range getPodAntiAffinityTerms(affinity.PodAntiAffinity) {
+		klog.Warningf("CandidatePod %s does not exist in nodeMap", klog.KObj(candidatePod))
-			namespaces := getNamespacesFromPodAffinityTerm(pod, &term)
+		return false
-			selector, err := metav1.LabelSelectorAsSelector(term.LabelSelector)
+	}
-			if err != nil {
+
-				klog.ErrorS(err, "Unable to convert LabelSelector into Selector")
+	affinity := candidatePod.Spec.Affinity
-				return false
+	if affinity == nil || affinity.PodAntiAffinity == nil {
-			}
+		return false
-			for namespace := range namespaces {
+	}
-				for _, existingPod := range pods[namespace] {
+
-					if existingPod.Name != pod.Name && podMatchesTermsNamespaceAndSelector(existingPod, namespaces, selector) {
+	for _, term := range GetPodAntiAffinityTerms(affinity.PodAntiAffinity) {
-						node, ok := nodeMap[pod.Spec.NodeName]
+		namespaces := GetNamespacesFromPodAffinityTerm(candidatePod, &term)
-						if !ok {
+		selector, err := metav1.LabelSelectorAsSelector(term.LabelSelector)
-							continue
+		if err != nil {
-						}
+			klog.ErrorS(err, "Unable to convert LabelSelector into Selector")
-						nodeHavingExistingPod, ok := nodeMap[existingPod.Spec.NodeName]
+			return false
-						if !ok {
+		}
-							continue
+
-						}
+		for namespace := range namespaces {
-						if hasSameLabelValue(node, nodeHavingExistingPod, term.TopologyKey) {
+			for _, assignedPod := range assignedPods[namespace] {
-							klog.V(1).InfoS("Found Pods matching PodAntiAffinity", "pod with anti-affinity", klog.KObj(pod))
+				if assignedPod.Name == candidatePod.Name || !PodMatchesTermsNamespaceAndSelector(assignedPod, namespaces, selector) {
-							return true
+					klog.V(4).InfoS("CandidatePod doesn't matches inter-pod anti-affinity rule of assigned pod on node", "candidatePod", klog.KObj(candidatePod), "assignedPod", klog.KObj(assignedPod))
-						}
+					continue
-					}
+				}
+
+				nodeHavingAssignedPod, ok := nodeMap[assignedPod.Spec.NodeName]
+				if !ok {
+					continue
+				}
+
+				if hasSameLabelValue(nodeHavingCandidatePod, nodeHavingAssignedPod, term.TopologyKey) {
+					klog.V(1).InfoS("CandidatePod matches inter-pod anti-affinity rule of assigned pod on node", "candidatePod", klog.KObj(candidatePod), "assignedPod", klog.KObj(assignedPod))
+					return true
 				}
 			}
 		}
 	}
+
 	return false
 }
 
-// getPodAntiAffinityTerms gets the antiaffinity terms for the given pod.
+// GetPodAntiAffinityTerms gets the antiaffinity terms for the given pod.
-func getPodAntiAffinityTerms(podAntiAffinity *v1.PodAntiAffinity) (terms []v1.PodAffinityTerm) {
+func GetPodAntiAffinityTerms(podAntiAffinity *v1.PodAntiAffinity) (terms []v1.PodAffinityTerm) {
 	if podAntiAffinity != nil {
 		if len(podAntiAffinity.RequiredDuringSchedulingIgnoredDuringExecution) != 0 {
 			terms = podAntiAffinity.RequiredDuringSchedulingIgnoredDuringExecution

pkg/utils/priority.go

@@ -4,42 +4,13 @@ import (
 	"context"
 	"fmt"
 
-	v1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/labels"
-	"k8s.io/apimachinery/pkg/util/sets"
 	clientset "k8s.io/client-go/kubernetes"
 	"sigs.k8s.io/descheduler/pkg/api"
 )
 
 const SystemCriticalPriority = 2 * int32(1000000000)
 
-// getNamespacesFromPodAffinityTerm returns a set of names
-// according to the namespaces indicated in podAffinityTerm.
-// If namespaces is empty it considers the given pod's namespace.
-func getNamespacesFromPodAffinityTerm(pod *v1.Pod, podAffinityTerm *v1.PodAffinityTerm) sets.Set[string] {
-	names := sets.New[string]()
-	if len(podAffinityTerm.Namespaces) == 0 {
-		names.Insert(pod.Namespace)
-	} else {
-		names.Insert(podAffinityTerm.Namespaces...)
-	}
-	return names
-}
-
-// podMatchesTermsNamespaceAndSelector returns true if the given <pod>
-// matches the namespace and selector defined by <affinityPod>`s <term>.
-func podMatchesTermsNamespaceAndSelector(pod *v1.Pod, namespaces sets.Set[string], selector labels.Selector) bool {
-	if !namespaces.Has(pod.Namespace) {
-		return false
-	}
-
-	if !selector.Matches(labels.Set(pod.Labels)) {
-		return false
-	}
-	return true
-}
-
 // GetPriorityFromPriorityClass gets priority from the given priority class.
 // If no priority class is provided, it will return SystemCriticalPriority by default.
 func GetPriorityFromPriorityClass(ctx context.Context, client clientset.Interface, name string) (int32, error) {