diff --git a/test/e2e/framework/volume/fixtures.go b/test/e2e/framework/volume/fixtures.go
index e64d7ad7fd1..9f0425ed55a 100644
--- a/test/e2e/framework/volume/fixtures.go
+++ b/test/e2e/framework/volume/fixtures.go
@@ -176,6 +176,24 @@ func NewNFSServerWithNodeName(ctx context.Context, cs clientset.Interface, names
 	return config, pod, host
 }
 
+// Restart the passed-in nfs-server by issuing a `rpc.nfsd 1` command in the
+// pod's (only) container. This command changes the number of nfs server threads from
+// (presumably) zero back to 1, and therefore allows nfs to open connections again.
+func RestartNFSServer(f *framework.Framework, serverPod *v1.Pod) {
+	const startcmd = "rpc.nfsd 1"
+	_, _, err := PodExec(f, serverPod, startcmd)
+	framework.ExpectNoError(err)
+}
+
+// Stop the passed-in nfs-server by issuing a `rpc.nfsd 0` command in the
+// pod's (only) container. This command changes the number of nfs server threads to 0,
+// thus closing all open nfs connections.
+func StopNFSServer(f *framework.Framework, serverPod *v1.Pod) {
+	const stopcmd = "rpc.nfsd 0 && for i in $(seq 200); do rpcinfo -p | grep -q nfs || break; sleep 1; done"
+	_, _, err := PodExec(f, serverPod, stopcmd)
+	framework.ExpectNoError(err)
+}
+
 // CreateStorageServer is a wrapper for startVolumeServer(). A storage server config is passed in, and a pod pointer
 // and ip address string are returned.
 // Note: Expect() is called so no error is returned.
diff --git a/test/e2e/node/kubelet.go b/test/e2e/node/kubelet.go
index 7757d268a94..c245697f30e 100644
--- a/test/e2e/node/kubelet.go
+++ b/test/e2e/node/kubelet.go
@@ -109,24 +109,6 @@ func waitTillNPodsRunningOnNodes(ctx context.Context, c clientset.Interface, nod
 	})
 }
 
-// Restart the passed-in nfs-server by issuing a `/usr/sbin/rpc.nfsd 1` command in the
-// pod's (only) container. This command changes the number of nfs server threads from
-// (presumably) zero back to 1, and therefore allows nfs to open connections again.
-func restartNfsServer(serverPod *v1.Pod) {
-	const startcmd = "/usr/sbin/rpc.nfsd 1"
-	ns := fmt.Sprintf("--namespace=%v", serverPod.Namespace)
-	e2ekubectl.RunKubectlOrDie(ns, "exec", ns, serverPod.Name, "--", "/bin/sh", "-c", startcmd)
-}
-
-// Stop the passed-in nfs-server by issuing a `/usr/sbin/rpc.nfsd 0` command in the
-// pod's (only) container. This command changes the number of nfs server threads to 0,
-// thus closing all open nfs connections.
-func stopNfsServer(serverPod *v1.Pod) {
-	const stopcmd = "/usr/sbin/rpc.nfsd 0"
-	ns := fmt.Sprintf("--namespace=%v", serverPod.Namespace)
-	e2ekubectl.RunKubectlOrDie(ns, "exec", ns, serverPod.Name, "--", "/bin/sh", "-c", stopcmd)
-}
-
 // Creates a pod that mounts an nfs volume that is served by the nfs-server pod. The container
 // will execute the passed in shell cmd. Waits for the pod to start.
 // Note: the nfs plugin is defined inline, no PV or PVC.
@@ -437,7 +419,7 @@ var _ = SIGDescribe("kubelet", func() {
 					pod = createPodUsingNfs(ctx, f, c, ns, nfsIP, t.podCmd)
 
 					ginkgo.By("Stop the NFS server")
-					stopNfsServer(nfsServerPod)
+					e2evolume.StopNFSServer(f, nfsServerPod)
 
 					ginkgo.By("Delete the pod mounted to the NFS volume -- expect failure")
 					err := e2epod.DeletePodWithWait(ctx, c, pod)
@@ -448,7 +430,7 @@ var _ = SIGDescribe("kubelet", func() {
 					checkPodCleanup(ctx, c, pod, false)
 
 					ginkgo.By("Restart the nfs server")
-					restartNfsServer(nfsServerPod)
+					e2evolume.RestartNFSServer(f, nfsServerPod)
 
 					ginkgo.By("Verify that the deleted client pod is now cleaned up")
 					checkPodCleanup(ctx, c, pod, true)
diff --git a/test/e2e_node/mirror_pod_test.go b/test/e2e_node/mirror_pod_test.go
index 70717c3a45b..7549640d5ad 100644
--- a/test/e2e_node/mirror_pod_test.go
+++ b/test/e2e_node/mirror_pod_test.go
@@ -231,7 +231,7 @@ var _ = SIGDescribe("MirrorPod", func() {
 			}
 
 			ginkgo.By("Stopping the NFS server")
-			stopNfsServer(f, nfsServerPod)
+			e2evolume.StopNFSServer(f, nfsServerPod)
 
 			ginkgo.By(fmt.Sprintf("Deleting the static nfs test pod: %s", staticPodName))
 			err = deleteStaticPod(podPath, staticPodName, ns)
@@ -243,7 +243,7 @@ var _ = SIGDescribe("MirrorPod", func() {
 			}, 5*time.Minute, 10*time.Second).Should(gomega.BeTrueBecause("pod volume should exist while nfs server is stopped"))
 
 			ginkgo.By("Start the NFS server")
-			restartNfsServer(f, nfsServerPod)
+			e2evolume.RestartNFSServer(f, nfsServerPod)
 
 			ginkgo.By("Waiting for the pod volume to deleted after the NFS server is started")
 			gomega.Eventually(func() bool {
@@ -286,25 +286,6 @@ func podVolumeDirectoryExists(uid types.UID) bool {
 	return podVolumeDirectoryExists
 }
 
-// Restart the passed-in nfs-server by issuing a `/usr/sbin/rpc.nfsd 1` command in the
-// pod's (only) container. This command changes the number of nfs server threads from
-// (presumably) zero back to 1, and therefore allows nfs to open connections again.
-func restartNfsServer(f *framework.Framework, serverPod *v1.Pod) {
-	const startcmd = "/usr/sbin/rpc.nfsd 1"
-	_, _, err := e2evolume.PodExec(f, serverPod, startcmd)
-	framework.ExpectNoError(err)
-
-}
-
-// Stop the passed-in nfs-server by issuing a `/usr/sbin/rpc.nfsd 0` command in the
-// pod's (only) container. This command changes the number of nfs server threads to 0,
-// thus closing all open nfs connections.
-func stopNfsServer(f *framework.Framework, serverPod *v1.Pod) {
-	const stopcmd = "rpc.nfsd 0 && for i in $(seq 200); do rpcinfo -p | grep -q nfs || break; sleep 1; done"
-	_, _, err := e2evolume.PodExec(f, serverPod, stopcmd)
-	framework.ExpectNoError(err)
-}
-
 func createStaticPodUsingNfs(nfsIP string, nodeName string, cmd string, dir string, name string, ns string) error {
 	ginkgo.By("create pod using nfs volume")