Merge pull request kubernetes#135 from vardhaman22/k8s-v1.27.14

[v1.27] Release v1.27.14
krunalhinguu · May 24, 2024 · 638a019 · 638a019
2 parents 194cb8b + 8070b86
commit 638a019
Show file tree

Hide file tree

Showing 33 changed files with 950 additions and 190 deletions.
diff --git a/CHANGELOG/CHANGELOG-1.27.md b/CHANGELOG/CHANGELOG-1.27.md
diff --git a/cluster/gce/config-default.sh b/cluster/gce/config-default.sh
@@ -86,7 +86,7 @@ fi
 # you are updating the os image versions, update this variable.
 # Also please update corresponding image for node e2e at:
 # https://github.com/kubernetes/kubernetes/blob/master/test/e2e_node/jenkins/image-config.yaml
-GCI_VERSION=${KUBE_GCI_VERSION:-cos-97-16919-103-16}
+GCI_VERSION=${KUBE_GCI_VERSION:-cos-109-17800-147-60}
 export MASTER_IMAGE=${KUBE_GCE_MASTER_IMAGE:-}
 export MASTER_IMAGE_PROJECT=${KUBE_GCE_MASTER_PROJECT:-cos-cloud}
 export NODE_IMAGE=${KUBE_GCE_NODE_IMAGE:-${GCI_VERSION}}

diff --git a/cluster/gce/config-test.sh b/cluster/gce/config-test.sh
@@ -99,7 +99,7 @@ ALLOWED_NOTREADY_NODES=${ALLOWED_NOTREADY_NODES:-$(($(get-num-nodes) / 100))}
 # you are updating the os image versions, update this variable.
 # Also please update corresponding image for node e2e at:
 # https://github.com/kubernetes/kubernetes/blob/master/test/e2e_node/jenkins/image-config.yaml
-GCI_VERSION=${KUBE_GCI_VERSION:-cos-97-16919-103-16}
+GCI_VERSION=${KUBE_GCI_VERSION:-cos-109-17800-147-60}
 export MASTER_IMAGE=${KUBE_GCE_MASTER_IMAGE:-}
 export MASTER_IMAGE_PROJECT=${KUBE_GCE_MASTER_PROJECT:-cos-cloud}
 export NODE_IMAGE=${KUBE_GCE_NODE_IMAGE:-${GCI_VERSION}}

diff --git a/pkg/scheduler/schedule_one.go b/pkg/scheduler/schedule_one.go
@@ -386,14 +386,17 @@ func (sched *Scheduler) schedulePod(ctx context.Context, fwk framework.Framework
 // Filters the nodes to find the ones that fit the pod based on the framework
 // filter plugins and filter extenders.
 func (sched *Scheduler) findNodesThatFitPod(ctx context.Context, fwk framework.Framework, state *framework.CycleState, pod *v1.Pod) ([]*v1.Node, framework.Diagnosis, error) {
-	diagnosis := framework.Diagnosis{
-		NodeToStatusMap:      make(framework.NodeToStatusMap),
-		UnschedulablePlugins: sets.NewString(),
-	}
-
 	allNodes, err := sched.nodeInfoSnapshot.NodeInfos().List()
 	if err != nil {
-		return nil, diagnosis, err
+		return nil, framework.Diagnosis{
+			NodeToStatusMap:      make(framework.NodeToStatusMap),
+			UnschedulablePlugins: sets.NewString(),
+		}, err
+	}
+
+	diagnosis := framework.Diagnosis{
+		NodeToStatusMap:      make(framework.NodeToStatusMap, len(allNodes)),
+		UnschedulablePlugins: sets.NewString(),
 	}
 	// Run "prefilter" plugins.
 	preRes, s := fwk.RunPreFilterPlugins(ctx, state, pod)
@@ -434,12 +437,14 @@ func (sched *Scheduler) findNodesThatFitPod(ctx context.Context, fwk framework.F
 	nodes := allNodes
 	if !preRes.AllNodes() {
 		nodes = make([]*framework.NodeInfo, 0, len(preRes.NodeNames))
-		for n := range preRes.NodeNames {
-			nInfo, err := sched.nodeInfoSnapshot.NodeInfos().Get(n)
-			if err != nil {
-				return nil, diagnosis, err
+		for _, n := range allNodes {
+			if !preRes.NodeNames.Has(n.Node().Name) {
+				// We consider Nodes that are filtered out by PreFilterResult as rejected via UnschedulableAndUnresolvable.
+				// We have to record them in NodeToStatusMap so that they won't be considered as candidates in the preemption.
+				diagnosis.NodeToStatusMap[n.Node().Name] = framework.NewStatus(framework.UnschedulableAndUnresolvable, "node is filtered out by the prefilter result")
+				continue
 			}
-			nodes = append(nodes, nInfo)
+			nodes = append(nodes, n)
 		}
 	}
 	feasibleNodes, err := sched.findNodesThatPassFilters(ctx, fwk, state, pod, diagnosis, nodes)

diff --git a/pkg/scheduler/schedule_one_test.go b/pkg/scheduler/schedule_one_test.go
@@ -1965,7 +1965,7 @@ func TestSchedulerSchedulePod(t *testing.T) {
 			nodes:              []string{"node1", "node2", "node3"},
 			pod:                st.MakePod().Name("test-prefilter").UID("test-prefilter").Obj(),
 			wantNodes:          sets.NewString("node2"),
-			wantEvaluatedNodes: pointer.Int32(1),
+			wantEvaluatedNodes: pointer.Int32(3),
 		},
 		{
 			name: "test prefilter plugin returning non-intersecting nodes",

diff --git a/pkg/scheduler/testing/wrappers.go b/pkg/scheduler/testing/wrappers.go
@@ -845,6 +845,20 @@ func (p *PersistentVolumeWrapper) HostPathVolumeSource(src *v1.HostPathVolumeSou
 	return p
 }
 
+// NodeAffinityIn creates a HARD node affinity (with the operator In)
+// and injects into the pv.
+func (p *PersistentVolumeWrapper) NodeAffinityIn(key string, vals []string) *PersistentVolumeWrapper {
+	if p.Spec.NodeAffinity == nil {
+		p.Spec.NodeAffinity = &v1.VolumeNodeAffinity{}
+	}
+	if p.Spec.NodeAffinity.Required == nil {
+		p.Spec.NodeAffinity.Required = &v1.NodeSelector{}
+	}
+	nodeSelector := MakeNodeSelector().In(key, vals).Obj()
+	p.Spec.NodeAffinity.Required.NodeSelectorTerms = append(p.Spec.NodeAffinity.Required.NodeSelectorTerms, nodeSelector.NodeSelectorTerms...)
+	return p
+}
+
 // ResourceClaimWrapper wraps a ResourceClaim inside.
 type ResourceClaimWrapper struct{ resourcev1alpha2.ResourceClaim }
 

diff --git a/staging/src/k8s.io/cloud-provider/cloud.go b/staging/src/k8s.io/cloud-provider/cloud.go
@@ -98,6 +98,8 @@ func DefaultLoadBalancerName(service *v1.Service) string {
 }
 
 // GetInstanceProviderID builds a ProviderID for a node in a cloud.
+// Note that if the instance does not exist, we must return ("", cloudprovider.InstanceNotFound)
+// cloudprovider.InstanceNotFound should NOT be returned for instances that exist but are stopped/sleeping
 func GetInstanceProviderID(ctx context.Context, cloud Interface, nodeName types.NodeName) (string, error) {
 	instances, ok := cloud.Instances()
 	if !ok {
@@ -108,8 +110,11 @@ func GetInstanceProviderID(ctx context.Context, cloud Interface, nodeName types.
 		if err == NotImplemented {
 			return "", err
 		}
+		if err == InstanceNotFound {
+			return "", err
+		}
 
-		return "", fmt.Errorf("failed to get instance ID from cloud provider: %v", err)
+		return "", fmt.Errorf("failed to get instance ID from cloud provider: %w", err)
 	}
 	return cloud.ProviderName() + "://" + instanceID, nil
 }

diff --git a/staging/src/k8s.io/cloud-provider/controllers/nodelifecycle/node_lifecycle_controller.go b/staging/src/k8s.io/cloud-provider/controllers/nodelifecycle/node_lifecycle_controller.go
@@ -152,7 +152,7 @@ func (c *CloudNodeLifecycleController) MonitorNodes(ctx context.Context) {
 
 		// At this point the node has NotReady status, we need to check if the node has been removed
 		// from the cloud provider. If node cannot be found in cloudprovider, then delete the node
-		exists, err := ensureNodeExistsByProviderID(ctx, c.cloud, node)
+		exists, err := c.ensureNodeExistsByProviderID(ctx, node)
 		if err != nil {
 			klog.Errorf("error checking if node %s exists: %v", node.Name, err)
 			continue
@@ -180,7 +180,7 @@ func (c *CloudNodeLifecycleController) MonitorNodes(ctx context.Context) {
 			// Node exists. We need to check this to get taint working in similar in all cloudproviders
 			// current problem is that shutdown nodes are not working in similar way ie. all cloudproviders
 			// does not delete node from kubernetes cluster when instance it is shutdown see issue #46442
-			shutdown, err := shutdownInCloudProvider(ctx, c.cloud, node)
+			shutdown, err := c.shutdownInCloudProvider(ctx, node)
 			if err != nil {
 				klog.Errorf("error checking if node %s is shutdown: %v", node.Name, err)
 			}
@@ -196,18 +196,49 @@ func (c *CloudNodeLifecycleController) MonitorNodes(ctx context.Context) {
 	}
 }
 
+// getProviderID returns the provider ID for the node. If Node CR has no provider ID,
+// it will be the one from the cloud provider.
+func (c *CloudNodeLifecycleController) getProviderID(ctx context.Context, node *v1.Node) (string, error) {
+	if node.Spec.ProviderID != "" {
+		return node.Spec.ProviderID, nil
+	}
+
+	if instanceV2, ok := c.cloud.InstancesV2(); ok {
+		metadata, err := instanceV2.InstanceMetadata(ctx, node)
+		if err != nil {
+			return "", err
+		}
+		return metadata.ProviderID, nil
+	}
+
+	providerID, err := cloudprovider.GetInstanceProviderID(ctx, c.cloud, types.NodeName(node.Name))
+	if err != nil {
+		return "", err
+	}
+
+	return providerID, nil
+}
+
 // shutdownInCloudProvider returns true if the node is shutdown on the cloud provider
-func shutdownInCloudProvider(ctx context.Context, cloud cloudprovider.Interface, node *v1.Node) (bool, error) {
-	if instanceV2, ok := cloud.InstancesV2(); ok {
+func (c *CloudNodeLifecycleController) shutdownInCloudProvider(ctx context.Context, node *v1.Node) (bool, error) {
+	if instanceV2, ok := c.cloud.InstancesV2(); ok {
 		return instanceV2.InstanceShutdown(ctx, node)
 	}
 
-	instances, ok := cloud.Instances()
+	instances, ok := c.cloud.Instances()
 	if !ok {
 		return false, errors.New("cloud provider does not support instances")
 	}
 
-	shutdown, err := instances.InstanceShutdownByProviderID(ctx, node.Spec.ProviderID)
+	providerID, err := c.getProviderID(ctx, node)
+	if err != nil {
+		if err == cloudprovider.InstanceNotFound {
+			return false, nil
+		}
+		return false, err
+	}
+
+	shutdown, err := instances.InstanceShutdownByProviderID(ctx, providerID)
 	if err == cloudprovider.NotImplemented {
 		return false, nil
 	}
@@ -216,32 +247,22 @@ func shutdownInCloudProvider(ctx context.Context, cloud cloudprovider.Interface,
 }
 
 // ensureNodeExistsByProviderID checks if the instance exists by the provider id,
-// If provider id in spec is empty it calls instanceId with node name to get provider id
-func ensureNodeExistsByProviderID(ctx context.Context, cloud cloudprovider.Interface, node *v1.Node) (bool, error) {
-	if instanceV2, ok := cloud.InstancesV2(); ok {
+func (c *CloudNodeLifecycleController) ensureNodeExistsByProviderID(ctx context.Context, node *v1.Node) (bool, error) {
+	if instanceV2, ok := c.cloud.InstancesV2(); ok {
 		return instanceV2.InstanceExists(ctx, node)
 	}
 
-	instances, ok := cloud.Instances()
+	instances, ok := c.cloud.Instances()
 	if !ok {
 		return false, errors.New("instances interface not supported in the cloud provider")
 	}
 
-	providerID := node.Spec.ProviderID
-	if providerID == "" {
-		var err error
-		providerID, err = instances.InstanceID(ctx, types.NodeName(node.Name))
-		if err != nil {
-			if err == cloudprovider.InstanceNotFound {
-				return false, nil
-			}
-			return false, err
-		}
-
-		if providerID == "" {
-			klog.Warningf("Cannot find valid providerID for node name %q, assuming non existence", node.Name)
+	providerID, err := c.getProviderID(ctx, node)
+	if err != nil {
+		if err == cloudprovider.InstanceNotFound {
 			return false, nil
 		}
+		return false, err
 	}
 
 	return instances.InstanceExistsByProviderID(ctx, providerID)