Spark-Submit提交作业过程

1. spark-submit脚本

exec $SPARK_HOME/bin/spark-class org.apache.spark.deploy.SparkSubmit "${ORIG_ARGS[@]}"

2. SparkSubmit中的main函数

def main(args: Array[String]): Unit = {
    val appArgs = new SparkSubmitArguments(args)
    if (appArgs.verbose) {
      printStream.println(appArgs)
    }
    appArgs.action match {
      case SparkSubmitAction.SUBMIT => submit(appArgs)
      case SparkSubmitAction.KILL => kill(appArgs)
      case SparkSubmitAction.REQUEST_STATUS => requestStatus(appArgs)
    }
  }

Internally, each RDD is characterized by five main properties:
 - A list of partitions

  //Implemented by subclasses to return the set of partitions in this RDD. This method will only be called once, so it is safe to implement a time-consuming computation in it.
  protected def getPartitions: Array[Partition]

- A function for computing each split

//Implemented by subclasses to compute a given partition.
def compute(split: Partition, context: TaskContext): Iterator[T]

 - A list of dependencies on other RDDs

//Implemented by subclasses to return how this RDD depends on parent RDDs. This method will only be called once, so it is safe to implement a time-consuming computation in it.
protected def getDependencies: Seq[Dependency[_]] = deps

 - Optionally, a Partitioner for key-value RDDs (e.g. to say that the RDD is hash-partitioned)

//Optionally overridden by subclasses to specify how they are partitioned. 
@transient val partitioner: Option[Partitioner] = None

 - Optionally, a list of preferred locations to compute each split on (e.g. block locations for an HDFS file)

//Optionally overridden by subclasses to specify placement preferences.
protected def getPreferredLocations(split: Partition): Seq[String] = Nil