Scala: Task not serializable in RDD map Caused by json4s “implicit val formats = DefaultFormats”

Question

The following program tries to invoke 3 functions for every ROW(in RDD map):

    import org.json4s._
    import org.json4s.jackson.JsonMethods._
    implicit val formats = DefaultFormats

    class TagCalculation extends Serializable {
    def test1(x: String) = x + " test1"
    def test2(x: String) = x + "test2" 
    def test3(x: String) = x + "test3" 
    def test5(arg1: java.lang.Integer, arg2: String, arg3: scala.collection.immutable.$colon$colon[Any]) = "test mix2"
  }
  val df = sqlContext.createDataFrame(Seq((1,"Android"), (2, "iPhone")))
  val get_test = new TagCalculation
  val field = Array("test1","test2","test3")

  val bb = df.rdd.map(row => {

    val reValue1 = "start"
    val ret = for(every <- field)
      yield {
        val test_para = Array(reValue1)
        val argtypes = test_para.map(_.getClass)
        val method4 = get_test.getClass.getMethod(every, argtypes: _*)

        val bbq = method4.invoke(get_test, test_para: _*)

        if (field.last == every)
            bbq
      }
    ret.last
  })

but some errors outputs:

org.apache.spark.SparkException: Task not serializable at org.apache.spark.util.ClosureCleaner$.ensureSerializable(ClosureCleaner.scala:304) at org.apache.spark.util.ClosureCleaner$.org$apache$spark$util$ClosureCleaner$$clean(ClosureCleaner.scala:294) at org.apache.spark.util.ClosureCleaner$.clean(ClosureCleaner.scala:122) at org.apache.spark.SparkContext.clean(SparkContext.scala:2032) at org.apache.spark.rdd.RDD$$anonfun$map$1.apply(RDD.scala:314) at org.apache.spark.rdd.RDD$$anonfun$map$1.apply(RDD.scala:313) at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:147) at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:108) at org.apache.spark.rdd.RDD.withScope(RDD.scala:306) at org.apache.spark.rdd.RDD.map(RDD.scala:313) ........ at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:205) at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:120) at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala) Caused by: java.io.NotSerializableException: org.json4s.DefaultFormats$

Any pointers?

It may be caused by "implicit val formats = DefaultFormats". But I need to extract value before "map".

Ramesh Maharjan Ramesh Maharjan · Accepted Answer · 2017-06-02T05:06:56

The problem is because you are defining TagCalculation class inside the calling class where you initialize and use the object. Just move it outside the calling class or make it a separate class and the issue with NotSerializableException should be solved.

Scala: Task not serializable in RDD map Caused by json4s “implicit val formats = DefaultFormats”

1 Answers