在看sparkContext之前,先回顾一下Scala的语法。Scala构造函数分主构造和辅构造函数,辅构造函数是关键字def+this定义的,而类中不在方法体也不在辅构造函数中的代码就是主构造函数,实例化对象的时候主构造函数都会被执行,例:
class person(name String,age Int){ println("主构造函数被调用") def this(name String,age Int){ //辅构造函数 this () //必须先调用主构造函数 this.name = name this.age = age } def introduce(){ println("name :" + name + "-age :" + age) } } val jack = new person("jack",2)
jack.introduce()
运行结果:
主构造函数被调用
name :jack-age :2
切入正题,看sparkContext的主构造函数比较重要的一些代码:
try{ ... // Create the Spark execution environment (cache, map output tracker, etc) _env = createSparkEnv(_conf, isLocal, listenerBus) SparkEnv.set(_env) ... // We need to register "HeartbeatReceiver" before "createTaskScheduler" because Executor will // retrieve "HeartbeatReceiver" in the constructor. (SPARK-6640) _heartbeatReceiver = env.rpcEnv.setupEndpoint( HeartbeatReceiver.ENDPOINT_NAME, new HeartbeatReceiver(this)) // Create and start the scheduler val (sched, ts) = SparkContext.createTaskScheduler(this, master, deployMode) _schedulerBackend = sched _taskScheduler = ts _dagScheduler = new DAGScheduler(this) _heartbeatReceiver.ask[Boolean](TaskSchedulerIsSet) // start TaskScheduler after taskScheduler sets DAGScheduler reference in DAGScheduler's // constructor _taskScheduler.start() }
首先:
_env = createSparkEnv(_conf, isLocal, listenerBus) SparkEnv.set(_env)
_heartbeatReceiver = env.rpcEnv.setupEndpoint( HeartbeatReceiver.ENDPOINT_NAME, new HeartbeatReceiver(this))
这里是在sparkContext中创建rpcEnv,并通过 setupEndpoint 向 rpcEnv 注册一个心跳的 Endpoint。
接着:
val (sched, ts) = SparkContext.createTaskScheduler(this, master, deployMode)
调的sparkContext自己的方法,创建taskScheduler,返回的是一个 (SchedulerBackend, TaskScheduler) 元组
private def createTaskScheduler( sc: SparkContext, master: String, deployMode: String): (SchedulerBackend, TaskScheduler) = { import SparkMasterRegex._ // When running locally, don't try to re-execute tasks on failure. val MAX_LOCAL_TASK_FAILURES = 1 master match { //... //standalone的提交模式 case SPARK_REGEX(sparkUrl) => val scheduler = new TaskSchedulerImpl(sc) val masterUrls = sparkUrl.split(",").map("spark://" + _) val backend = new StandaloneSchedulerBackend(scheduler, sc, masterUrls) //调用初始化方法 scheduler.initialize(backend) (backend, scheduler) } //... }
方法内部根据master参数判断不同的提交模式,创建不同的(SchedulerBackend, TaskScheduler) ,拿standalon模式举例,根据入参创建TaskSchedulerImpl和StandalonSchedulerBackend,再调用TaskSchedulerImpl的初始化方法,最后返回一个元组。
scheduler.initialize(backend),其实就是根据不同的schedulingMode创建不同的schedulableBuilder,它就是对Scheduleable tree的封装,负责对taskSet的调度。
def initialize(backend: SchedulerBackend) { this.backend = backend schedulableBuilder = { schedulingMode match { case SchedulingMode.FIFO => new FIFOSchedulableBuilder(rootPool) case SchedulingMode.FAIR => new FairSchedulableBuilder(rootPool, conf) case _ => throw new IllegalArgumentException(s"Unsupported $SCHEDULER_MODE_PROPERTY: " + s"$schedulingMode") } } schedulableBuilder.buildPools() }
接着下面两行代码:
_dagScheduler = new DAGScheduler(this)
创建DAG有向无环图,实现类面向stage的调度机制的高层次调度层,他会为每个stage计算DAG(有向无环图),追踪RDD和stage的输出是否被物化(写入磁盘或内存),并且寻找一个最少消耗的调度机制来运行job。它会将stage作为taskSets提交到底层的TaskSchedulerImpl上来在集群运行。除了处理stage的DAG,它还负责决定运行每个task的最佳位置,基于当前的缓存状态,将最佳位置提交给底层的TaskSchedulerImpl,此外,他会处理由于每个shuffle输出文件导致的失败,在这种情况下旧的stage可能会被重新提交。一个stage内部的失败,如果不是由于shuffle文件丢失导致的失败,会被taskScheduler处理,它会多次重试每个task,还不行才会取消整个stage。
_heartbeatReceiver.ask[Boolean](TaskSchedulerIsSet)
在上面创建好了TaskScheduler和SchedulerBackend后,告诉 HeartbeatReceiver(心跳) 的监听端。
最后:
_taskScheduler.start()
在TaskSchedulerImpl的start()方法中调的是SchedulerBackend的start()方法,所以start()方法运行的是这段:
override def start() { super.start() // SPARK-21159. The scheduler backend should only try to connect to the launcher when in client // mode. In cluster mode, the code that submits the application to the Master needs to connect // to the launcher instead. if (sc.deployMode == "client") { launcherBackend.connect() } //参数设置 val appDesc = ApplicationDescription(sc.appName, maxCores, sc.executorMemory, command, webUrl, sc.eventLogDir, sc.eventLogCodec, coresPerExecutor, initialExecutorLimit) client = new StandaloneAppClient(sc.env.rpcEnv, masters, appDesc, this, conf) client.start() launcherBackend.setState(SparkAppHandle.State.SUBMITTED) waitForRegistration() launcherBackend.setState(SparkAppHandle.State.RUNNING) }
这里创建了两个对象:AppliactionDescription和AppClient,AppliactionDescription顾名思义就是对Application的描述类,比如它需要的资源等;AppClient负责负责为application与spark集群通信。SchedulerBackend的start()最终调用了AppClient的start(),代码如下:
def start() { // Just launch an rpcEndpoint; it will call back into the listener. endpoint.set(rpcEnv.setupEndpoint("AppClient", new ClientEndpoint(rpcEnv))) }
启动一个rpcEndPoint并回调给监听器,RPC原理可看这篇 https://www.cnblogs.com/superhedantou/p/7570692.html
最后画个大概流程图