test: Drain tool E2Es, UTs

test/e2e/setup.sh

@@ -39,12 +39,12 @@ REGIONS=("" "" "eastasia")
 AKS_NODE_REGIONS=("westus" "northeurope" "eastasia")
 # The SKUs that should be set on each node of the respective clusters; if the AKS property
 # provider is used. See the AKS documentation for specifics.
-# 
+#
 # Note that this is for information only; kind nodes always use the same fixed setup
 # (total/allocatable capacity = host capacity).
 AKS_NODE_SKUS=("Standard_A4_v2" "Standard_B4ms" "Standard_D8s_v5" "Standard_E16_v5" "Standard_M16ms")
 AKS_SKU_COUNT=${#AKS_NODE_SKUS[@]}
-# The number of clusters that has pre-defined configuration for testing purposes. 
+# The number of clusters that has pre-defined configuration for testing purposes.
 RESERVED_CLUSTER_COUNT=${MEMBER_CLUSTER_COUNT}
 
 # Create the kind clusters
@@ -87,7 +87,7 @@ then
             k=$(( RANDOM % AKS_SKU_COUNT ))
             kubectl label node "${NODES[$j]}" beta.kubernetes.io/instance-type=${AKS_NODE_SKUS[$k]}
         done
-    done 
+    done
 fi
 
 # Build the Fleet agent images
@@ -207,3 +207,13 @@ do
     fi
 done
 
+# Create tools directory if it doesn't exist
+mkdir -p ../../hack/tools/bin
+
+# Build drain binary
+echo "Building drain binary..."
+go build -o ../../hack/tools/bin/kubectl-draincluster ../../tools/draincluster
+
+# Build uncordon binary
+echo "Building uncordon binary..."
+go build -o ../../hack/tools/bin/kubectl-uncordoncluster ../../tools/uncordoncluster

test/e2e/setup_test.go

@@ -19,8 +19,10 @@ package e2e
 import (
 	"context"
 	"flag"
+	"fmt"
 	"log"
 	"os"
+	"path/filepath"
 	"sync"
 	"testing"
 	"time"
@@ -163,6 +165,11 @@ var (
 	}
 )
 
+var (
+	drainBinaryPath    = filepath.Join("../../", "hack", "tools", "bin", "kubectl-draincluster")
+	uncordonBinaryPath = filepath.Join("../../", "hack", "tools", "bin", "kubectl-uncordoncluster")
+)
+
 var (
 	isAzurePropertyProviderEnabled = (os.Getenv(propertyProviderEnvVarName) == azurePropertyProviderEnvVarValue)
 
@@ -340,6 +347,12 @@ func beforeSuiteForAllProcesses() {
 		for i := range allMemberClusters {
 			allMemberClusterNames = append(allMemberClusterNames, allMemberClusters[i].ClusterName)
 		}
+
+		// Check if drain cluster and uncordon cluster binaries exist.
+		_, err := os.Stat(drainBinaryPath)
+		Expect(os.IsNotExist(err)).To(BeFalse(), fmt.Sprintf("drain binary not found at %s", drainBinaryPath))
+		_, err = os.Stat(uncordonBinaryPath)
+		Expect(os.IsNotExist(err)).To(BeFalse(), fmt.Sprintf("uncordon binary not found at %s", uncordonBinaryPath))
 	})
 }
 

tools/draincluster/README.md

@@ -3,11 +3,11 @@
 1. Build the binary for the `draincluster` tool by running the following command in the root directory of the fleet repo:
 
 ```bash
-go build -o ./hack/tools/bin/kubectl-draincluster ./tools/draincluster/main.go
+go build -o ./hack/tools/bin/kubectl-draincluster ./tools/draincluster/
 ```
 
 2. Copy the binary to a directory in your `PATH` so that it can be run as a kubectl plugin. For example, you can move it to
-`/usr/local/bin`:
+   `/usr/local/bin`:
 
 ```bash
 sudo cp ./hack/tools/bin/kubectl-draincluster /usr/local/bin/
@@ -33,13 +33,13 @@ The following compatible plugins are available:
 /usr/local/bin/kubectl-draincluster
 ```
 
-please refer to the [kubectl plugin documentation](https://kubernetes.io/docs/tasks/extend-kubectl/kubectl-plugins/) for 
+please refer to the [kubectl plugin documentation](https://kubernetes.io/docs/tasks/extend-kubectl/kubectl-plugins/) for
 more information.
 
 # Drain Member Cluster connected to a fleet
 
-After following the steps above to build the `draincluster` tool as a kubectl plugin, you can use it to remove all 
-resources propagated to the member cluster from the hub cluster by any `Placement` resource. This is useful when you 
+After following the steps above to build the `draincluster` tool as a kubectl plugin, you can use it to remove all
+resources propagated to the member cluster from the hub cluster by any `Placement` resource. This is useful when you
 want to temporarily move all workloads off a member cluster in preparation for an event like upgrade or reconfiguration.
 
 The `draincluster` tool can be used to drain a member cluster by running the following command:
@@ -68,8 +68,8 @@ CURRENT   NAME               CLUSTER            AUTHINFO
 
 Here you can see that the context of the hub cluster is called `hub` under the `NAME` column.
 
-The command adds a `Taint` to the `MemberCluster` resource of the member cluster to prevent any new resources from being 
-propagated to the member cluster. Then it creates `Eviction` objects for all the `Placement` objects that have propagated 
+The command adds a `Taint` to the `MemberCluster` resource of the member cluster to prevent any new resources from being
+propagated to the member cluster. Then it creates `Eviction` objects for all the `Placement` objects that have propagated
 resources to the member cluster.
 
 >> **Note**: The `draincluster` tool is a best-effort mechanism at the moment, so once the command is run successfully

tools/draincluster/drain/drain.go → tools/draincluster/drain.go

@@ -1,9 +1,20 @@
 /*
-Copyright (c) Microsoft Corporation.
-Licensed under the MIT license.
+Copyright 2025 The KubeFleet Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
 */
 
-package drain
+package main
 
 import (
 	"context"
@@ -32,31 +43,31 @@ const (
 	resourceIdentifierKeyFormat = "%s/%s/%s/%s/%s"
 )
 
-type Helper struct {
-	HubClient   client.Client
-	ClusterName string
+type helper struct {
+	hubClient   client.Client
+	clusterName string
 }
 
-func (h *Helper) Drain(ctx context.Context) (bool, error) {
+func (h *helper) Drain(ctx context.Context) (bool, error) {
 	if err := h.cordon(ctx); err != nil {
-		return false, fmt.Errorf("failed to cordon member cluster %s: %w", h.ClusterName, err)
+		return false, fmt.Errorf("failed to cordon member cluster %s: %w", h.clusterName, err)
 	}
-	log.Printf("Successfully cordoned member cluster %s by adding cordon taint", h.ClusterName)
+	log.Printf("Successfully cordoned member cluster %s by adding cordon taint", h.clusterName)
 
 	crpNameMap, err := h.fetchClusterResourcePlacementNamesToEvict(ctx)
 	if err != nil {
 		return false, err
 	}
 
 	if len(crpNameMap) == 0 {
-		log.Printf("There are currently no resources propagated to %s from fleet using ClusterResourcePlacement resources", h.ClusterName)
+		log.Printf("There are currently no resources propagated to %s from fleet using ClusterResourcePlacement resources", h.clusterName)
 		return true, nil
 	}
 
 	isDrainSuccessful := true
 	// create eviction objects for all <crpName, targetCluster>.
 	for crpName := range crpNameMap {
-		evictionName, err := generateDrainEvictionName(crpName, h.ClusterName)
+		evictionName, err := generateDrainEvictionName(crpName, h.clusterName)
 		if err != nil {
 			return false, err
 		}
@@ -70,27 +81,29 @@ func (h *Helper) Drain(ctx context.Context) (bool, error) {
 				},
 				Spec: placementv1beta1.PlacementEvictionSpec{
 					PlacementName: crpName,
-					ClusterName:   h.ClusterName,
+					ClusterName:   h.clusterName,
 				},
 			}
-			return h.HubClient.Create(ctx, &eviction)
+			return h.hubClient.Create(ctx, &eviction)
 		})
 
 		if err != nil {
-			return false, fmt.Errorf("failed to create eviction for CRP %s: %w", crpName, err)
+			return false, fmt.Errorf("failed to create eviction %s for CRP %s targeting member cluster %s: %w", evictionName, crpName, h.clusterName, err)
 		}
 
+		log.Printf("Created eviction %s for CRP %s targeting member cluster %s", evictionName, crpName, h.clusterName)
+
 		// wait until evictions reach a terminal state.
 		var eviction placementv1beta1.ClusterResourcePlacementEviction
 		err = wait.ExponentialBackoffWithContext(ctx, retry.DefaultBackoff, func(ctx context.Context) (bool, error) {
-			if err := h.HubClient.Get(ctx, types.NamespacedName{Name: evictionName}, &eviction); err != nil {
-				return false, fmt.Errorf("failed to get eviction %s: %w", evictionName, err)
+			if err := h.hubClient.Get(ctx, types.NamespacedName{Name: evictionName}, &eviction); err != nil {
+				return false, fmt.Errorf("failed to get eviction %s for CRP %s targeting member cluster %s: %w", evictionName, crpName, h.clusterName, err)
 			}
 			return evictionutils.IsEvictionInTerminalState(&eviction), nil
 		})
 
 		if err != nil {
-			return false, fmt.Errorf("failed to wait for evictions to reach terminal state: %w", err)
+			return false, fmt.Errorf("failed to wait for eviction %s for CRP %s targeting member cluster %s to reach terminal state: %w", evictionName, crpName, h.clusterName, err)
 		}
 
 		// TODO: add safeguards to check if eviction conditions are set to unknown.
@@ -100,36 +113,36 @@ func (h *Helper) Drain(ctx context.Context) (bool, error) {
 			if validCondition.Reason == condition.EvictionInvalidMissingCRPMessage ||
 				validCondition.Reason == condition.EvictionInvalidDeletingCRPMessage ||
 				validCondition.Reason == condition.EvictionInvalidMissingCRBMessage {
-				log.Printf("eviction %s is invalid with reason %s for CRP %s, but drain will succeed", evictionName, validCondition.Reason, crpName)
+				log.Printf("eviction %s is invalid with reason %s for CRP %s targeting member cluster %s, but drain will succeed", evictionName, validCondition.Reason, crpName, h.clusterName)
 				continue
 			}
 		}
 		executedCondition := eviction.GetCondition(string(placementv1beta1.PlacementEvictionConditionTypeExecuted))
 		if executedCondition == nil || executedCondition.Status == metav1.ConditionFalse {
 			isDrainSuccessful = false
-			log.Printf("eviction %s was not executed successfully for CRP %s", evictionName, crpName)
+			log.Printf("eviction %s was not executed successfully for CRP %s targeting member cluster %s", evictionName, crpName, h.clusterName)
 			continue
 		}
-		log.Printf("eviction %s was executed successfully for CRP %s", evictionName, crpName)
+		log.Printf("eviction %s was executed successfully for CRP %s targeting member cluster %s", evictionName, crpName, h.clusterName)
 		// log each cluster scoped resource evicted for CRP.
 		clusterScopedResourceIdentifiers, err := h.collectClusterScopedResourcesSelectedByCRP(ctx, crpName)
 		if err != nil {
 			log.Printf("failed to collect cluster scoped resources selected by CRP %s: %v", crpName, err)
 			continue
 		}
 		for _, resourceIdentifier := range clusterScopedResourceIdentifiers {
-			log.Printf("evicted resource %s propagated by CRP %s", generateResourceIdentifierKey(resourceIdentifier), crpName)
+			log.Printf("evicted resource %s propagated by CRP %s targeting member cluster %s", generateResourceIdentifierKey(resourceIdentifier), crpName, h.clusterName)
 		}
 	}
 
 	return isDrainSuccessful, nil
 }
 
-func (h *Helper) cordon(ctx context.Context) error {
+func (h *helper) cordon(ctx context.Context) error {
 	// add taint to member cluster to ensure resources aren't scheduled on it.
 	return retry.RetryOnConflict(retry.DefaultRetry, func() error {
 		var mc clusterv1beta1.MemberCluster
-		if err := h.HubClient.Get(ctx, types.NamespacedName{Name: h.ClusterName}, &mc); err != nil {
+		if err := h.hubClient.Get(ctx, types.NamespacedName{Name: h.clusterName}, &mc); err != nil {
 			return err
 		}
 
@@ -143,21 +156,21 @@ func (h *Helper) cordon(ctx context.Context) error {
 		// add taint to member cluster to cordon.
 		mc.Spec.Taints = append(mc.Spec.Taints, toolsutils.CordonTaint)
 
-		return h.HubClient.Update(ctx, &mc)
+		return h.hubClient.Update(ctx, &mc)
 	})
 }
 
-func (h *Helper) fetchClusterResourcePlacementNamesToEvict(ctx context.Context) (map[string]bool, error) {
+func (h *helper) fetchClusterResourcePlacementNamesToEvict(ctx context.Context) (map[string]bool, error) {
 	var crbList placementv1beta1.ClusterResourceBindingList
-	if err := h.HubClient.List(ctx, &crbList); err != nil {
+	if err := h.hubClient.List(ctx, &crbList); err != nil {
 		return map[string]bool{}, fmt.Errorf("failed to list cluster resource bindings: %w", err)
 	}
 
 	crpNameMap := make(map[string]bool)
 	// find all unique CRP names for which eviction needs to occur.
 	for i := range crbList.Items {
 		crb := crbList.Items[i]
-		if crb.Spec.TargetCluster == h.ClusterName && crb.DeletionTimestamp == nil {
+		if crb.Spec.TargetCluster == h.clusterName && crb.DeletionTimestamp == nil {
 			crpName, ok := crb.GetLabels()[placementv1beta1.CRPTrackingLabel]
 			if !ok {
 				return map[string]bool{}, fmt.Errorf("failed to get CRP name from binding %s", crb.Name)
@@ -169,9 +182,9 @@ func (h *Helper) fetchClusterResourcePlacementNamesToEvict(ctx context.Context)
 	return crpNameMap, nil
 }
 
-func (h *Helper) collectClusterScopedResourcesSelectedByCRP(ctx context.Context, crpName string) ([]placementv1beta1.ResourceIdentifier, error) {
+func (h *helper) collectClusterScopedResourcesSelectedByCRP(ctx context.Context, crpName string) ([]placementv1beta1.ResourceIdentifier, error) {
 	var crp placementv1beta1.ClusterResourcePlacement
-	if err := h.HubClient.Get(ctx, types.NamespacedName{Name: crpName}, &crp); err != nil {
+	if err := h.hubClient.Get(ctx, types.NamespacedName{Name: crpName}, &crp); err != nil {
 		return nil, fmt.Errorf("failed to get ClusterResourcePlacement %s: %w", crpName, err)
 	}
 
@@ -188,7 +201,8 @@ func (h *Helper) collectClusterScopedResourcesSelectedByCRP(ctx context.Context,
 func generateDrainEvictionName(crpName, targetCluster string) (string, error) {
 	evictionName := fmt.Sprintf(drainEvictionNameFormat, crpName, targetCluster, uuid.NewUUID()[:uuidLength])
 
-	if errs := validation.IsQualifiedName(evictionName); len(errs) != 0 {
+	// check to see if eviction name is a valid DNS1123 subdomain name https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#dns-subdomain-names.
+	if errs := validation.IsDNS1123Subdomain(evictionName); len(errs) != 0 {
 		return "", fmt.Errorf("failed to format a qualified name for drain eviction object with CRP name %s, cluster name %s: %v", crpName, targetCluster, errs)
 	}
 	return evictionName, nil
@@ -198,6 +212,9 @@ func generateResourceIdentifierKey(r placementv1beta1.ResourceIdentifier) string
 	if len(r.Group) == 0 && len(r.Namespace) == 0 {
 		return fmt.Sprintf(resourceIdentifierKeyFormat, "''", r.Version, r.Kind, "''", r.Name)
 	}
+	if len(r.Group) == 0 {
+		return fmt.Sprintf(resourceIdentifierKeyFormat, "''", r.Version, r.Kind, r.Namespace, r.Name)
+	}
 	if len(r.Namespace) == 0 {
 		return fmt.Sprintf(resourceIdentifierKeyFormat, r.Group, r.Version, r.Kind, "''", r.Name)
 	}