Spark writing to hdfs not working with the saveAsNewAPIHadoopFile method

问题

I am using Spark 1.1.0 on CDH 5.2.0 and am trying to make sure that I can read from and write to hdfs.

I quickly realized that the .textFile and .saveAsTextFile call the old api and do not seem to be compatible with our hdfs version.

  def testHDFSReadOld(sc: SparkContext, readFile: String){
    //THIS WILL FAIL WITH
    //(TID 0, dl1rhd416.internal.edmunds.com): java.lang.IllegalStateException: unread block data
    //java.io.ObjectInputStream$BlockDataInputStream.setBlockDataMode(ObjectInputStream.java:2420)

    sc.textFile(readFile).take(2).foreach(println)
  }

  def testHDFSWriteOld(sc: SparkContext, writeFile: String){
    //THIS WILL FAIL WITH
    //(TID 0, dl1rhd416.internal.edmunds.com): java.lang.IllegalStateException: unread block data
    //java.io.ObjectInputStream$BlockDataInputStream.setBlockDataMode(ObjectInputStream.java:2420)

    sc.parallelize(List("THIS","ISCOOL")).saveAsTextFile(writeFile)
  }

Moving over to the newAPI methods fixed reading from hdfs!

  def testHDFSReadNew(sc: SparkContext, readFile: String){
    //THIS WORKS
    sc.newAPIHadoopFile(readFile, classOf[TextInputFormat], classOf[LongWritable],
      classOf[Text],sc.hadoopConfiguration).map{
      case (x:LongWritable, y: Text) => y.toString
    }.take(2).foreach(println)
  }

So it seemed like I was making progress. Writing no longer exited with hard error like it did above, instead it appeared to be working. The only problem is that apart from a lonely SUCCESS flag file in the directory, there was nothing. Further baffling, was that the logs showed that data was being written to the _temporary directory. It just seems like the file committer never realized it needed to move the files from the _temporary directory to the output directory.

  def testHDFSWriteNew(sc: SparkContext, writeFile: String){
    /*This will have an error message of:
    INFO ConnectionManager: Removing SendingConnection to ConnectionManagerId(dl1rhd400.internal.edmunds.com,35927)
    14/11/21 02:02:27 INFO ConnectionManager: Key not valid ? sun.nio.ch.SelectionKeyImpl@2281f1b2
      14/11/21 02:02:27 INFO ConnectionManager: key already cancelled ? sun.nio.ch.SelectionKeyImpl@2281f1b2
      java.nio.channels.CancelledKeyException
    at org.apache.spark.network.ConnectionManager.run(ConnectionManager.scala:386)
    at org.apache.spark.network.ConnectionManager$$anon$4.run(ConnectionManager.scala:139)

    However lately it hasn't even had errors, symptoms are no part files in the directory but a success flag is there
    */
    val conf = sc.hadoopConfiguration
    conf.set("mapreduce.task.files.preserve.failedtasks", "true")
    conf.set("mapred.output.dir", writeFile)
    sc.parallelize(List("THIS","ISCOOL")).map(x => (NullWritable.get, new Text(x)))
      .saveAsNewAPIHadoopFile(writeFile, classOf[NullWritable], classOf[Text], classOf[TextOutputFormat[NullWritable, Text]], conf)

  }

When I run locally and specify hdfs paths, the files appear fine in hdfs. This only happens when I run on our spark standalone cluster.

I submit a job as follows: spark-submit --deploy-mode client --master spark://sparkmaster --class driverclass driverjar

回答1:

Can you try with following code ?

import org.apache.hadoop.io._
import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
val nums = sc.makeRDD(1 to 3).map(x => (new IntWritable(x), new Text("a" * x)))
nums.saveAsNewAPIHadoopFile[TextOutputFormat[IntWritable, Text]]("/data/newAPIHadoopFile")

Following code also worked for me.

val x = sc.parallelize(List("THIS","ISCOOL")).map(x => (NullWritable.get, new Text(x)))
x.saveAsNewAPIHadoopFile("/data/nullwritable", classOf[NullWritable], classOf[Text], classOf[TextOutputFormat[NullWritable, Text]], sc.hadoopConfiguration)

[root@sparkmaster ~]# hadoop fs -cat /data/nullwritable/*

15/08/20 02:09:19 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable

来源：https://stackoverflow.com/questions/27072911/spark-writing-to-hdfs-not-working-with-the-saveasnewapihadoopfile-method