diff --git a/go.mod b/go.mod
index dc2cf83e4..d605ffef0 100644
--- a/go.mod
+++ b/go.mod
@@ -15,5 +15,6 @@ require (
 	k8s.io/code-generator v0.19.3
 	k8s.io/component-base v0.19.3
 	k8s.io/klog/v2 v2.2.0
+	k8s.io/utils v0.0.0-20200729134348-d5654de09c73
 	sigs.k8s.io/controller-runtime v0.6.4
 )
diff --git a/pkg/controllers/policy/policy_controller.go b/pkg/controllers/policy/policy_controller.go
index c91479878..d984a9da9 100644
--- a/pkg/controllers/policy/policy_controller.go
+++ b/pkg/controllers/policy/policy_controller.go
@@ -209,6 +209,9 @@ func (c *PropagationPolicyController) claimResources(owner string, workloads []*
 			return err
 		}
 		workloadLabel := workload.GetLabels()
+		if workloadLabel == nil {
+			workloadLabel = make(map[string]string, 1)
+		}
 		workloadLabel[util.PolicyClaimLabel] = owner
 		workload.SetLabels(workloadLabel)
 		_, err = c.DynamicClient.Resource(dynamicResource).Namespace(workload.GetNamespace()).Update(context.TODO(), workload, metav1.UpdateOptions{})
diff --git a/test/e2e/policy_test.go b/test/e2e/policy_test.go
new file mode 100644
index 000000000..d99c1fb55
--- /dev/null
+++ b/test/e2e/policy_test.go
@@ -0,0 +1,86 @@
+package e2e
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/onsi/ginkgo"
+	"github.com/onsi/gomega"
+	"k8s.io/apimachinery/pkg/api/errors"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/util/rand"
+	"k8s.io/apimachinery/pkg/util/wait"
+
+	"github.com/karmada-io/karmada/test/helper"
+)
+
+var _ = ginkgo.Describe("[propagation policy] propagation policy functionality testing", func() {
+	deploymentName := rand.String(6)
+	deployment := helper.NewDeployment(testNamespace, deploymentName)
+	var err error
+
+	ginkgo.BeforeEach(func() {
+		_, err = kubeClient.AppsV1().Deployments(testNamespace).Create(context.TODO(), deployment, metav1.CreateOptions{})
+		gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+	})
+
+	ginkgo.AfterEach(func() {
+		err = kubeClient.CoreV1().Namespaces().Delete(context.TODO(), testNamespace, metav1.DeleteOptions{})
+		gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+	})
+
+	// propagate single resource to two explicit clusters.
+	ginkgo.Context("single resource propagation testing", func() {
+		ginkgo.It("propagate deployment", func() {
+			policyName := rand.String(6)
+			policyNamespace := testNamespace // keep policy in the same namespace with the resource
+			policy := helper.NewPolicyWithSingleDeployment(policyNamespace, policyName, deployment, memberClusterNames)
+
+			ginkgo.By(fmt.Sprintf("creating policy: %s/%s", policyNamespace, policyName), func() {
+				_, err = karmadaClient.PropagationstrategyV1alpha1().PropagationPolicies(policyNamespace).Create(context.TODO(), policy, metav1.CreateOptions{})
+				gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+			})
+			ginkgo.By("check if resource appear in member clusters", func() {
+				for _, cluster := range memberClusters {
+					clusterClient := getClusterClient(cluster.Name)
+					gomega.Expect(clusterClient).ShouldNot(gomega.BeNil())
+
+					err = wait.Poll(pollInterval, pollTimeout, func() (done bool, err error) {
+						_, err = clusterClient.AppsV1().Deployments(deployment.Namespace).Get(context.TODO(), deployment.Name, metav1.GetOptions{})
+						if err != nil {
+							return false, err
+						}
+						return true, nil
+					})
+					gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+				}
+			})
+
+			ginkgo.By(fmt.Sprintf("deleting policy: %s/%s", policyNamespace, policyName), func() {
+				err = karmadaClient.PropagationstrategyV1alpha1().PropagationPolicies(policyNamespace).Delete(context.TODO(), policyName, metav1.DeleteOptions{})
+				gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+			})
+			ginkgo.By("check if resource disappear from member clusters", func() {
+				for _, cluster := range memberClusters {
+					clusterClient := getClusterClient(cluster.Name)
+					gomega.Expect(clusterClient).ShouldNot(gomega.BeNil())
+
+					err = wait.Poll(pollInterval, pollTimeout, func() (done bool, err error) {
+						_, err = clusterClient.AppsV1().Deployments(deployment.Namespace).Get(context.TODO(), deployment.Name, metav1.GetOptions{})
+						if err != nil {
+							if errors.IsNotFound(err) {
+								return true, nil
+							}
+						}
+						return false, nil
+					})
+					gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+				}
+			})
+		})
+	})
+
+	// propagate two resource to two explicit clusters.
+	ginkgo.Context("multiple resource propagation testing", func() {
+	})
+})
diff --git a/test/e2e/suite_test.go b/test/e2e/suite_test.go
index 2084f5c34..e878130ce 100644
--- a/test/e2e/suite_test.go
+++ b/test/e2e/suite_test.go
@@ -11,13 +11,16 @@ import (
 	"github.com/onsi/gomega"
 
 	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/util/rand"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/rest"
 	"k8s.io/client-go/tools/clientcmd"
 	"k8s.io/klog/v2"
 
+	clusterapi "github.com/karmada-io/karmada/pkg/apis/membercluster/v1alpha1"
 	karmada "github.com/karmada-io/karmada/pkg/generated/clientset/versioned"
 	"github.com/karmada-io/karmada/pkg/util"
+	"github.com/karmada-io/karmada/test/helper"
 )
 
 const (
@@ -26,15 +29,24 @@ const (
 	// TestSuiteTeardownTimeOut defines the time after which the suite tear down times out.
 	TestSuiteTeardownTimeOut = 300 * time.Second
 
+	// pollInterval defines the interval time for a poll operation.
+	pollInterval = 5 * time.Second
+	// pollTimeout defines the time after which the poll operation times out.
+	pollTimeout = 60 * time.Second
+
 	// MinimumMemberCluster represents the minimum number of member clusters to run E2E test.
 	MinimumMemberCluster = 2
 )
 
 var (
-	kubeconfig    string
-	restConfig    *rest.Config
-	kubeClient    kubernetes.Interface
-	karmadaClient karmada.Interface
+	kubeconfig           string
+	restConfig           *rest.Config
+	kubeClient           kubernetes.Interface
+	karmadaClient        karmada.Interface
+	memberClusters       []*clusterapi.MemberCluster
+	memberClusterNames   []string
+	memberClusterClients []*util.ClusterClient
+	testNamespace        = fmt.Sprintf("karmada-e2e-%s", rand.String(3))
 )
 
 func TestE2E(t *testing.T) {
@@ -56,35 +68,110 @@ var _ = ginkgo.BeforeSuite(func() {
 	karmadaClient, err = karmada.NewForConfig(restConfig)
 	gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
 
-	meetRequirement, err := isMemberClusterMeetRequirements(karmadaClient)
+	memberClusters, err = fetchMemberClusters(karmadaClient)
+	gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+
+	var meetRequirement bool
+	meetRequirement, err = isMemberClusterMeetRequirements(memberClusters)
 	gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
 	gomega.Expect(meetRequirement).Should(gomega.BeTrue())
+
+	for _, cluster := range memberClusters {
+		memberClusterNames = append(memberClusterNames, cluster.Name)
+
+		memberClusterClient, err := util.NewClusterClientSet(cluster, kubeClient)
+		gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
+		memberClusterClients = append(memberClusterClients, memberClusterClient)
+	}
+	gomega.Expect(memberClusterNames).Should(gomega.HaveLen(len(memberClusters)))
+
+	err = setupTestNamespace(testNamespace, kubeClient, memberClusterClients)
+	gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
 }, TestSuiteSetupTimeOut.Seconds())
 
 var _ = ginkgo.AfterSuite(func() {
-	// suite tear down, such as cleanup karmada environment.
+	// cleanup all namespaces we created both in control plane and member clusters.
+	// It will not return error even if there is no such namespace in there that may happen in case setup failed.
+	err := cleanupTestNamespace(testNamespace, kubeClient, memberClusterClients)
+	gomega.Expect(err).ShouldNot(gomega.HaveOccurred())
 }, TestSuiteTeardownTimeOut.Seconds())
 
-// isMemberClusterMeetRequirements checks if current environment meet the requirements of E2E.
-func isMemberClusterMeetRequirements(client karmada.Interface) (bool, error) {
-	// list all member cluster we have
-	clusters, err := client.MemberclusterV1alpha1().MemberClusters().List(context.TODO(), v1.ListOptions{})
+// fetchMemberClusters will fetch all member clusters we have.
+func fetchMemberClusters(client karmada.Interface) ([]*clusterapi.MemberCluster, error) {
+	clusterList, err := client.MemberclusterV1alpha1().MemberClusters().List(context.TODO(), v1.ListOptions{})
 	if err != nil {
-		return false, err
+		return nil, err
 	}
 
+	clusters := make([]*clusterapi.MemberCluster, 0, len(clusterList.Items))
+	for _, cluster := range clusterList.Items {
+		pinedCluster := cluster
+		clusters = append(clusters, &pinedCluster)
+	}
+
+	return clusters, nil
+}
+
+// isMemberClusterMeetRequirements checks if current environment meet the requirements of E2E.
+func isMemberClusterMeetRequirements(clusters []*clusterapi.MemberCluster) (bool, error) {
 	// check if member cluster number meets requirements
-	if len(clusters.Items) < MinimumMemberCluster {
-		return false, fmt.Errorf("needs at lease %d member cluster to run, but got: %d", MinimumMemberCluster, len(clusters.Items))
+	if len(clusters) < MinimumMemberCluster {
+		return false, fmt.Errorf("needs at lease %d member cluster to run, but got: %d", MinimumMemberCluster, len(clusters))
 	}
 
 	// check if all member cluster status is ready
-	for _, cluster := range clusters.Items {
-		if !util.IsMemberClusterReady(&cluster) {
+	for _, cluster := range clusters {
+		if !util.IsMemberClusterReady(cluster) {
 			return false, fmt.Errorf("cluster %s not ready", cluster.GetName())
 		}
 	}
 
-	klog.Infof("Got %d member cluster and all in ready state.", len(clusters.Items))
+	klog.Infof("Got %d member cluster and all in ready state.", len(clusters))
 	return true, nil
 }
+
+// setupTestNamespace will create a namespace in control plane and all member clusters, most of cases will run against it.
+// The reason why we need a separated namespace is it will make it easier to cleanup resources deployed by the testing.
+func setupTestNamespace(namespace string, kubeClient kubernetes.Interface, memberClusterClients []*util.ClusterClient) error {
+	namespaceObj := helper.NewNamespace(namespace)
+	_, err := util.CreateNamespace(kubeClient, namespaceObj)
+	if err != nil {
+		return err
+	}
+
+	for _, clusterClient := range memberClusterClients {
+		_, err = util.CreateNamespace(clusterClient.KubeClient, namespaceObj)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// cleanupTestNamespace will remove the namespace we setup before for the whole testing.
+func cleanupTestNamespace(namespace string, kubeClient kubernetes.Interface, memberClusterClients []*util.ClusterClient) error {
+	err := util.DeleteNamespace(kubeClient, namespace)
+	if err != nil {
+		return err
+	}
+
+	for _, clusterClient := range memberClusterClients {
+		err = util.DeleteNamespace(clusterClient.KubeClient, namespace)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func getClusterClient(clusterName string) kubernetes.Interface {
+	for _, client := range memberClusterClients {
+		if client.ClusterName == clusterName {
+			return client.KubeClient
+		}
+	}
+
+	return nil
+}
diff --git a/test/helper/deployment.go b/test/helper/deployment.go
new file mode 100644
index 000000000..8e6f56441
--- /dev/null
+++ b/test/helper/deployment.go
@@ -0,0 +1,41 @@
+package helper
+
+import (
+	appsv1 "k8s.io/api/apps/v1"
+	corev1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/utils/pointer"
+)
+
+// NewDeployment will build a deployment object.
+func NewDeployment(namespace string, name string) *appsv1.Deployment {
+	podLabels := map[string]string{"app": "nginx"}
+
+	return &appsv1.Deployment{
+		TypeMeta: metav1.TypeMeta{
+			APIVersion: "apps/v1",
+			Kind:       "Deployment",
+		},
+		ObjectMeta: metav1.ObjectMeta{
+			Namespace: namespace,
+			Name:      name,
+		},
+		Spec: appsv1.DeploymentSpec{
+			Replicas: pointer.Int32Ptr(3),
+			Selector: &metav1.LabelSelector{
+				MatchLabels: podLabels,
+			},
+			Template: corev1.PodTemplateSpec{
+				ObjectMeta: metav1.ObjectMeta{
+					Labels: podLabels,
+				},
+				Spec: corev1.PodSpec{
+					Containers: []corev1.Container{{
+						Name:  "nginx",
+						Image: "nginx:1.19.0",
+					}},
+				},
+			},
+		},
+	}
+}
diff --git a/test/helper/namespace.go b/test/helper/namespace.go
new file mode 100644
index 000000000..0ce364302
--- /dev/null
+++ b/test/helper/namespace.go
@@ -0,0 +1,15 @@
+package helper
+
+import (
+	corev1 "k8s.io/api/core/v1"
+	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+)
+
+// NewNamespace will build a Namespace object.
+func NewNamespace(namespace string) *corev1.Namespace {
+	return &corev1.Namespace{
+		ObjectMeta: v1.ObjectMeta{
+			Name: namespace,
+		},
+	}
+}
diff --git a/test/helper/propagationpolicy.go b/test/helper/propagationpolicy.go
new file mode 100644
index 000000000..96bdcb2ac
--- /dev/null
+++ b/test/helper/propagationpolicy.go
@@ -0,0 +1,33 @@
+package helper
+
+import (
+	appsv1 "k8s.io/api/apps/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+
+	propagationapi "github.com/karmada-io/karmada/pkg/apis/propagationstrategy/v1alpha1"
+)
+
+// NewPolicyWithSingleDeployment will build a PropagationPolicy object.
+func NewPolicyWithSingleDeployment(namespace string, name string, deployment *appsv1.Deployment, clusters []string) *propagationapi.PropagationPolicy {
+	return &propagationapi.PropagationPolicy{
+		ObjectMeta: metav1.ObjectMeta{
+			Namespace: namespace,
+			Name:      name,
+		},
+		Spec: propagationapi.PropagationSpec{
+			ResourceSelectors: []propagationapi.ResourceSelector{
+				{
+					APIVersion: deployment.APIVersion,
+					Kind:       deployment.Kind,
+					Names:      []string{deployment.Name},
+					Namespaces: []string{deployment.Namespace},
+				},
+			},
+			Placement: propagationapi.Placement{
+				ClusterAffinity: &propagationapi.ClusterAffinity{
+					ClusterNames: clusters,
+				},
+			},
+		},
+	}
+}
diff --git a/vendor/modules.txt b/vendor/modules.txt
index a365c291b..670e0a12a 100644
--- a/vendor/modules.txt
+++ b/vendor/modules.txt
@@ -794,6 +794,7 @@ k8s.io/kube-openapi/pkg/util
 k8s.io/kube-openapi/pkg/util/proto
 k8s.io/kube-openapi/pkg/util/sets
 # k8s.io/utils v0.0.0-20200729134348-d5654de09c73
+## explicit
 k8s.io/utils/buffer
 k8s.io/utils/integer
 k8s.io/utils/net