kubelet: monitor the health of pleg

PLEG is reponsible for listing the pods running on the node. If it's hung
due to non-responsive container runtime or internal bugs, we should restart
kubelet.

pkg/kubelet/server/server.go

@@ -166,6 +166,7 @@ type HostInterface interface {
 	DockerImagesFsInfo() (cadvisorapiv2.FsInfo, error)
 	RootFsInfo() (cadvisorapiv2.FsInfo, error)
 	ListVolumesForPod(podUID types.UID) (map[string]volume.Volume, bool)
+	PLEGHealthCheck() (bool, error)
 }
 
 // NewServer initializes and configures a kubelet.Server object to handle HTTP requests.
@@ -223,6 +224,7 @@ func (s *Server) InstallDefaultHandlers() {
 	healthz.InstallHandler(s.restfulCont,
 		healthz.PingHealthz,
 		healthz.NamedCheck("syncloop", s.syncLoopHealthCheck),
+		healthz.NamedCheck("pleg", s.plegHealthCheck),
 	)
 	var ws *restful.WebService
 	ws = new(restful.WebService)
@@ -385,6 +387,14 @@ func (s *Server) syncLoopHealthCheck(req *http.Request) error {
 	return nil
 }
 
+// Checks if pleg, which lists pods periodically, is healthy.
+func (s *Server) plegHealthCheck(req *http.Request) error {
+	if ok, err := s.host.PLEGHealthCheck(); !ok {
+		return fmt.Errorf("PLEG took longer than expected: %v", err)
+	}
+	return nil
+}
+
 // getContainerLogs handles containerLogs request against the Kubelet
 func (s *Server) getContainerLogs(request *restful.Request, response *restful.Response) {
 	podNamespace := request.PathParameter("podNamespace")