SparkCore之RDD中的函数传递

在实际开发中我们往往需要自己定义一些对于RDD的操作，那么此时需要主要的是，初始化工作是在Driver端进行的，而实际运行程序是在Executor端进行的，这就涉及到了跨进程通信，是需要序列化的。下面我们看几个例子：

1 传递一个方法

1．创建一个类

class Search(query:String){

 

//过滤出包含字符串的数据

  def isMatch(s: String): Boolean = {

    s.contains(query)

  }

 

//过滤出包含字符串的RDD

  def getMatch1 (rdd: RDD[String]): RDD[String] = {

    rdd.filter(isMatch)

  }

 

  //过滤出包含字符串的RDD

  def getMatche2(rdd: RDD[String]): RDD[String] = {

    rdd.filter(x => x.contains(query))

  }

 

}

2．创建Spark主程序

object SeriTest {

 

  def main(args: Array[String]): Unit = {

 

    //1.初始化配置信息及SparkContext

    val sparkConf: SparkConf = new SparkConf().setAppName("WordCount").setMaster("local[*]")

    val sc = new SparkContext(sparkConf)

 

//2.创建一个RDD

    val rdd: RDD[String] = sc.parallelize(Array("hadoop", "spark", "hive", "atguigu"))

 

//3.创建一个Search对象

    val search = new Search()

 

//4.运用第一个过滤函数并打印结果

    val match1: RDD[String] = search.getMatche1(rdd)

    match1.collect().foreach(println)

    }

}

3．运行程序

Exception in thread "main" org.apache.spark.SparkException: Task not serializable

    at org.apache.spark.util.ClosureCleaner$.ensureSerializable(ClosureCleaner.scala:298)

    at org.apache.spark.util.ClosureCleaner$.org$apache$spark$util$ClosureCleaner$$clean(ClosureCleaner.scala:288)

    at org.apache.spark.util.ClosureCleaner$.clean(ClosureCleaner.scala:108)

    at org.apache.spark.SparkContext.clean(SparkContext.scala:2101)

    at org.apache.spark.rdd.RDD$$anonfun$filter$1.apply(RDD.scala:387)

    at org.apache.spark.rdd.RDD$$anonfun$filter$1.apply(RDD.scala:386)

    at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:151)

    at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:112)

    at org.apache.spark.rdd.RDD.withScope(RDD.scala:362)

    at org.apache.spark.rdd.RDD.filter(RDD.scala:386)

    at com.atguigu.Search.getMatche1(SeriTest.scala:39)

    at com.atguigu.SeriTest$.main(SeriTest.scala:18)

    at com.atguigu.SeriTest.main(SeriTest.scala)

Caused by: java.io.NotSerializableException: com.atguigu.Search

4．问题说明

//过滤出包含字符串的RDD

  def getMatch1 (rdd: RDD[String]): RDD[String] = {

    rdd.filter(isMatch)

  }

在这个方法中所调用的方法isMatch()是定义在Search这个类中的，实际上调用的是this. isMatch()，this表示Search这个类的对象，程序在运行过程中需要将Search对象序列化以后传递到Executor端。

5．解决方案

使类继承scala.Serializable即可。

class Search() extends Serializable{...}

 

2 传递一个属性

1．创建Spark主程序

object TransmitTest {

 

  def main(args: Array[String]): Unit = {

 

    //1.初始化配置信息及SparkContext

    val sparkConf: SparkConf = new SparkConf().setAppName("WordCount").setMaster("local[*]")

    val sc = new SparkContext(sparkConf)

 

//2.创建一个RDD

    val rdd: RDD[String] = sc.parallelize(Array("hadoop", "spark", "hive", "atguigu"))

 

//3.创建一个Search对象

    val search = new Search()

 

//4.运用第一个过滤函数并打印结果

    val match1: RDD[String] = search.getMatche2(rdd)

    match1.collect().foreach(println)

    }

}

2．运行程序

Exception in thread "main" org.apache.spark.SparkException: Task not serializable

    at org.apache.spark.util.ClosureCleaner$.ensureSerializable(ClosureCleaner.scala:298)

    at org.apache.spark.util.ClosureCleaner$.org$apache$spark$util$ClosureCleaner$$clean(ClosureCleaner.scala:288)

    at org.apache.spark.util.ClosureCleaner$.clean(ClosureCleaner.scala:108)

    at org.apache.spark.SparkContext.clean(SparkContext.scala:2101)

    at org.apache.spark.rdd.RDD$$anonfun$filter$1.apply(RDD.scala:387)

    at org.apache.spark.rdd.RDD$$anonfun$filter$1.apply(RDD.scala:386)

    at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:151)

    at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:112)

    at org.apache.spark.rdd.RDD.withScope(RDD.scala:362)

    at org.apache.spark.rdd.RDD.filter(RDD.scala:386)

    at com.atguigu.Search.getMatche1(SeriTest.scala:39)

    at com.atguigu.SeriTest$.main(SeriTest.scala:18)

    at com.atguigu.SeriTest.main(SeriTest.scala)

Caused by: java.io.NotSerializableException: com.atguigu.Search

3．问题说明

  //过滤出包含字符串的RDD

  def getMatche2(rdd: RDD[String]): RDD[String] = {

    rdd.filter(x => x.contains(query))

  }

在这个方法中所调用的方法query是定义在Search这个类中的字段，实际上调用的是this. query，this表示Search这个类的对象，程序在运行过程中需要将Search对象序列化以后传递到Executor端。

4．解决方案

1）使类继承scala.Serializable即可。

class Search() extends Serializable{...}

2）将类变量query赋值给局部变量

修改getMatche2为

  //过滤出包含字符串的RDD

  def getMatche2(rdd: RDD[String]): RDD[String] = {

    val query_ : String = this.query//将类变量赋值给局部变量

    rdd.filter(x => x.contains(query_))

  }