Spark广播/序列化错误

Question

我已经为一个Mahout作业的Spark版本创建了一个名为“item similarity”的CLI驱动程序，其中有几个测试都可以在本地[4] Spark standalone上正常工作。该代码甚至可以读取和写入集群HDFS。但切换到群集Spark有一个似乎与广播和/或序列化相关的问题。

该代码使用HashBiMap，这是一个Guava Java的东西。为每个Mahout drm（一个分布式矩阵）创建了两个，用于双向行和列ID查找。它们被创建一次，然后广播到处访问。

当我在群集Spark上运行这个时，我得到以下错误。有一次我们使用HashMaps，他们似乎在群集上工作。所以我怀疑有关HashBiMap的一些问题。我也怀疑它可能与广播中的序列化有关。这是一段代码和错误。

// create BiMaps for bi-directional lookup of ID by either Mahout ID or external ID 
// broadcast them for access in distributed processes, so they are not recalculated in every task. 
// rowIDDictionary is a HashBiMap[String, Int] 
val rowIDDictionary = asOrderedDictionary(rowIDs) // this creates the HashBiMap in a non-dsitributed manner 
val rowIDDictionary_bcast = mc.broadcast(rowIDDictionary) 

val columnIDDictionary = asOrderedDictionary(columnIDs)) // this creates the HashBiMap in a non-dsitributed manner 
val columnIDDictionary_bcast = mc.broadcast(columnIDDictionary) 

val indexedInteractions = 
    interactions.map { case (rowID, columnID) => //<<<<<<<<<<< this is the stage being submitted before the error 
    val rowIndex = rowIDDictionary_bcast.value.get(rowID).get 
    val columnIndex = columnIDDictionary_bcast.value.get(columnID).get 

    rowIndex -> columnIndex 
    } 

错误似乎发生在访问_bcast vals时执行interact.map时。任何想法从哪里开始寻找这个？

14/06/26 11:23:36 INFO scheduler.DAGScheduler: Submitting Stage 9 (MappedRDD[17] at map at TextDelimitedReaderWriter.scala:83), which has no missing parents 
14/06/26 11:23:36 INFO scheduler.DAGScheduler: Submitting 2 missing tasks from Stage 9 (MappedRDD[17] at map at TextDelimitedReaderWriter.scala:83) 
14/06/26 11:23:36 INFO scheduler.TaskSchedulerImpl: Adding task set 9.0 with 2 tasks 
14/06/26 11:23:36 INFO scheduler.TaskSetManager: Starting task 9.0:0 as TID 16 on executor 0: occam4 (PROCESS_LOCAL) 
14/06/26 11:23:36 INFO scheduler.TaskSetManager: Serialized task 9.0:0 as 2418 bytes in 0 ms 
14/06/26 11:23:36 INFO scheduler.TaskSetManager: Starting task 9.0:1 as TID 17 on executor 0: occam4 (PROCESS_LOCAL) 
14/06/26 11:23:36 INFO scheduler.TaskSetManager: Serialized task 9.0:1 as 2440 bytes in 0 ms 
14/06/26 11:23:36 WARN scheduler.TaskSetManager: Lost TID 16 (task 9.0:0) 
14/06/26 11:23:36 WARN scheduler.TaskSetManager: Loss was due to java.lang.NullPointerException 
java.lang.NullPointerException 
    at com.google.common.collect.HashBiMap.seekByKey(HashBiMap.java:180) 
    at com.google.common.collect.HashBiMap.put(HashBiMap.java:230) 
    at com.google.common.collect.HashBiMap.put(HashBiMap.java:218) 
    at com.esotericsoftware.kryo.serializers.MapSerializer.read(MapSerializer.java:135) 
    at com.esotericsoftware.kryo.serializers.MapSerializer.read(MapSerializer.java:17) 
    at com.esotericsoftware.kryo.Kryo.readClassAndObject(Kryo.java:729) 
    at org.apache.spark.serializer.KryoDeserializationStream.readObject(KryoSerializer.scala:102) 
    at org.apache.spark.broadcast.HttpBroadcast$.read(HttpBroadcast.scala:165) 
    at org.apache.spark.broadcast.HttpBroadcast.readObject(HttpBroadcast.scala:56) 
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) 
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39) 
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25) 
    at java.lang.reflect.Method.invoke(Method.java:597) 
    at java.io.ObjectStreamClass.invokeReadObject(ObjectStreamClass.java:969) 
    at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1871) 
    at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1775) 
    at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1327) 
    at java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1969) 
    at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1893) 
    at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1775) 
    at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1327) 
    at java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1969) 
    at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1893) 
    at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1775) 
    at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1327) 
    at java.io.ObjectInputStream.readObject(ObjectInputStream.java:349) 
    at org.apache.spark.serializer.JavaDeserializationStream.readObject(JavaSerializer.scala:40) 
    at org.apache.spark.scheduler.ShuffleMapTask$.deserializeInfo(ShuffleMapTask.scala:69) 
    at org.apache.spark.scheduler.ShuffleMapTask.readExternal(ShuffleMapTask.scala:138) 
    at java.io.ObjectInputStream.readExternalData(ObjectInputStream.java:1814) 
    at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1773) 
    at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1327) 
    at java.io.ObjectInputStream.readObject(ObjectInputStream.java:349) 
    at org.apache.spark.serializer.JavaDeserializationStream.readObject(JavaSerializer.scala:40) 
    at org.apache.spark.serializer.JavaSerializerInstance.deserialize(JavaSerializer.scala:62) 
    at org.apache.spark.executor.Executor$TaskRunner$$anonfun$run$1.apply$mcV$sp(Executor.scala:193) 
    at org.apache.spark.deploy.SparkHadoopUtil$$anon$1.run(SparkHadoopUtil.scala:42) 
    at org.apache.spark.deploy.SparkHadoopUtil$$anon$1.run(SparkHadoopUtil.scala:41) 
    at java.security.AccessController.doPrivileged(Native Method) 
    at javax.security.auth.Subject.doAs(Subject.java:396) 
    at   org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1190) 
    at org.apache.spark.deploy.SparkHadoopUtil.runAsUser(SparkHadoopUtil.scala:41) 
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:176) 
    at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:895) 
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918) 
    at java.lang.Thread.run(Thread.java:662) 

Answer 1

它看起来像你使用kryo序列化，你是否也在你的本地测试中使用它？ 如果用于HashBiMap的kryo序列化不成功，您可能希望显式注册类kryo以使用Java serilization。