diff --git a/pkg/kubelet/container/helpers.go b/pkg/kubelet/container/helpers.go
index b0b9d0c84bb..2831a72e9b8 100644
--- a/pkg/kubelet/container/helpers.go
+++ b/pkg/kubelet/container/helpers.go
@@ -83,30 +83,6 @@ func ShouldContainerBeRestarted(container *api.Container, pod *api.Pod, podStatu
 	return true
 }
 
-// TODO(random-liu): Convert PodStatus to running Pod, should be deprecated soon
-func ConvertPodStatusToRunningPod(podStatus *PodStatus) Pod {
-	runningPod := Pod{
-		ID:        podStatus.ID,
-		Name:      podStatus.Name,
-		Namespace: podStatus.Namespace,
-	}
-	for _, containerStatus := range podStatus.ContainerStatuses {
-		if containerStatus.State != ContainerStateRunning {
-			continue
-		}
-		container := &Container{
-			ID:      containerStatus.ID,
-			Name:    containerStatus.Name,
-			Image:   containerStatus.Image,
-			ImageID: containerStatus.ImageID,
-			Hash:    containerStatus.Hash,
-			State:   containerStatus.State,
-		}
-		runningPod.Containers = append(runningPod.Containers, container)
-	}
-	return runningPod
-}
-
 // HashContainer returns the hash of the container. It is used to compare
 // the running container with its desired spec.
 func HashContainer(container *api.Container) uint64 {
diff --git a/pkg/kubelet/dockertools/docker_manager.go b/pkg/kubelet/dockertools/docker_manager.go
index b8f892842c8..89c7f12d203 100644
--- a/pkg/kubelet/dockertools/docker_manager.go
+++ b/pkg/kubelet/dockertools/docker_manager.go
@@ -48,6 +48,7 @@ import (
 	kubecontainer "k8s.io/kubernetes/pkg/kubelet/container"
 	"k8s.io/kubernetes/pkg/kubelet/events"
 	"k8s.io/kubernetes/pkg/kubelet/images"
+	"k8s.io/kubernetes/pkg/kubelet/kuberuntime"
 	"k8s.io/kubernetes/pkg/kubelet/lifecycle"
 	"k8s.io/kubernetes/pkg/kubelet/metrics"
 	"k8s.io/kubernetes/pkg/kubelet/network"
@@ -2051,7 +2052,7 @@ func (dm *DockerManager) SyncPod(pod *api.Pod, _ api.PodStatus, podStatus *kubec
 
 		// Killing phase: if we want to start new infra container, or nothing is running kill everything (including infra container)
 		// TODO(random-liu): We'll use pod status directly in the future
-		killResult := dm.killPodWithSyncResult(pod, kubecontainer.ConvertPodStatusToRunningPod(podStatus), nil)
+		killResult := dm.killPodWithSyncResult(pod, kuberuntime.ConvertPodStatusToRunningPod(dm.Type(), podStatus), nil)
 		result.AddPodSyncResult(killResult)
 		if killResult.Error() != nil {
 			return
diff --git a/pkg/kubelet/kubelet.go b/pkg/kubelet/kubelet.go
index 6bb73a86512..4fc348515bf 100644
--- a/pkg/kubelet/kubelet.go
+++ b/pkg/kubelet/kubelet.go
@@ -1675,7 +1675,7 @@ func (kl *Kubelet) killPod(pod *api.Pod, runningPod *kubecontainer.Pod, status *
 	if runningPod != nil {
 		p = *runningPod
 	} else if status != nil {
-		p = kubecontainer.ConvertPodStatusToRunningPod(status)
+		p = kuberuntime.ConvertPodStatusToRunningPod(kl.GetRuntime().Type(), status)
 	}
 	return kl.containerRuntime.KillPod(pod, p, gracePeriodOverride)
 }
diff --git a/pkg/kubelet/kuberuntime/helpers.go b/pkg/kubelet/kuberuntime/helpers.go
index c7909b49d78..e2a06e57af6 100644
--- a/pkg/kubelet/kuberuntime/helpers.go
+++ b/pkg/kubelet/kuberuntime/helpers.go
@@ -92,6 +92,39 @@ func sandboxToKubeContainerState(state runtimeApi.PodSandBoxState) kubecontainer
 	return kubecontainer.ContainerStateUnknown
 }
 
+// TODO(random-liu): Convert PodStatus to running Pod, should be deprecated soon
+func ConvertPodStatusToRunningPod(runtimeName string, podStatus *kubecontainer.PodStatus) kubecontainer.Pod {
+	runningPod := kubecontainer.Pod{
+		ID:        podStatus.ID,
+		Name:      podStatus.Name,
+		Namespace: podStatus.Namespace,
+	}
+	for _, containerStatus := range podStatus.ContainerStatuses {
+		if containerStatus.State != kubecontainer.ContainerStateRunning {
+			continue
+		}
+		container := &kubecontainer.Container{
+			ID:      containerStatus.ID,
+			Name:    containerStatus.Name,
+			Image:   containerStatus.Image,
+			ImageID: containerStatus.ImageID,
+			Hash:    containerStatus.Hash,
+			State:   containerStatus.State,
+		}
+		runningPod.Containers = append(runningPod.Containers, container)
+	}
+
+	// Need to place a sandbox in the Pod as well.
+	for _, sandbox := range podStatus.SandboxStatuses {
+		runningPod.Sandboxes = append(runningPod.Sandboxes, &kubecontainer.Container{
+			ID:    kubecontainer.ContainerID{Type: runtimeName, ID: *sandbox.Id},
+			State: sandboxToKubeContainerState(*sandbox.State),
+		})
+	}
+
+	return runningPod
+}
+
 // toRuntimeProtocol converts api.Protocol to runtimeApi.Protocol.
 func toRuntimeProtocol(protocol api.Protocol) runtimeApi.Protocol {
 	switch protocol {
diff --git a/pkg/kubelet/kuberuntime/kuberuntime_manager.go b/pkg/kubelet/kuberuntime/kuberuntime_manager.go
index 33729240af5..8a9bb7f7d79 100644
--- a/pkg/kubelet/kuberuntime/kuberuntime_manager.go
+++ b/pkg/kubelet/kuberuntime/kuberuntime_manager.go
@@ -483,7 +483,7 @@ func (m *kubeGenericRuntimeManager) SyncPod(pod *api.Pod, _ api.PodStatus, podSt
 			glog.V(4).Infof("Stopping PodSandbox for %q, will start new one", format.Pod(pod))
 		}
 
-		killResult := m.killPodWithSyncResult(pod, kubecontainer.ConvertPodStatusToRunningPod(podStatus), nil)
+		killResult := m.killPodWithSyncResult(pod, ConvertPodStatusToRunningPod(m.runtimeName, podStatus), nil)
 		result.AddPodSyncResult(killResult)
 		if killResult.Error() != nil {
 			glog.Errorf("killPodWithSyncResult failed: %v", killResult.Error())
diff --git a/pkg/kubelet/rkt/rkt.go b/pkg/kubelet/rkt/rkt.go
index ca106d012ee..7c0ce8056e3 100644
--- a/pkg/kubelet/rkt/rkt.go
+++ b/pkg/kubelet/rkt/rkt.go
@@ -47,6 +47,7 @@ import (
 	kubecontainer "k8s.io/kubernetes/pkg/kubelet/container"
 	"k8s.io/kubernetes/pkg/kubelet/events"
 	"k8s.io/kubernetes/pkg/kubelet/images"
+	"k8s.io/kubernetes/pkg/kubelet/kuberuntime"
 	"k8s.io/kubernetes/pkg/kubelet/leaky"
 	"k8s.io/kubernetes/pkg/kubelet/lifecycle"
 	"k8s.io/kubernetes/pkg/kubelet/network"
@@ -1710,7 +1711,7 @@ func (r *Runtime) SyncPod(pod *api.Pod, podStatus api.PodStatus, internalPodStat
 	// TODO: (random-liu) Stop using running pod in SyncPod()
 	// TODO: (random-liu) Rename podStatus to apiPodStatus, rename internalPodStatus to podStatus, and use new pod status as much as possible,
 	// we may stop using apiPodStatus someday.
-	runningPod := kubecontainer.ConvertPodStatusToRunningPod(internalPodStatus)
+	runningPod := kuberuntime.ConvertPodStatusToRunningPod(r.Type(), internalPodStatus)
 	// Add references to all containers.
 	unidentifiedContainers := make(map[kubecontainer.ContainerID]*kubecontainer.Container)
 	for _, c := range runningPod.Containers {