StandaloneAppClient是什麼?這個很容易搞混淆。其實StandaloneAppClient不是SparkApplication,它主要是用在ScheduleBackend中的。app
獨立集羣環境中,ScheduleBackend是用的StandaloneScheduleBackend,它繼承了CoarseGrainedSchedulerBackend類。ide
StandaloneScheduleBackend裏面用了一個叫StandaloneAppClient的類,這個StandaloneAppClient很具備迷惑性,其實它的主要功能是替換CoarseGrainedSchedulerBackend的資源申請的方法,改成向Master申請資源,咱們看看相關代碼片斷就好了。spa
先看他啓動的時候:code
private def tryRegisterAllMasters(): Array[JFuture[_]] = { for (masterAddress <- masterRpcAddresses) yield { registerMasterThreadPool.submit(new Runnable { override def run(): Unit = try { if (registered.get) { return } logInfo("Connecting to master " + masterAddress.toSparkURL + "...") val masterRef = rpcEnv.setupEndpointRef(masterAddress, Master.ENDPOINT_NAME) masterRef.send(RegisterApplication(appDescription, self)) } catch { } }) } }
向Master發送RegisterApplication消息,將本appDesc註冊給Master,這個和DriverDescription註冊到Master是有點區別的。繼承
再好比資源申請的代碼:ip
def requestTotalExecutors(requestedTotal: Int): Future[Boolean] = { if (endpoint.get != null && appId.get != null) { endpoint.get.ask[Boolean](RequestExecutors(appId.get, requestedTotal)) } else { logWarning("Attempted to request executors before driver fully initialized.") Future.successful(false) } }
就是向Master發送RequestExecutor消息申請Executor資源。資源
這裏爲啥要註冊Application到Master呢?主要是當Master失效或者Master更改時,能通知到Application,這樣就能從新鏈接新的Master了,從新運行spark程序。不然就很脆弱,很容易崩潰,這是個人理解哦,不必定正確,^~^rpc