Spark：当加入2个大型DF时，大小超过Integer.MAX_VALUE

时间：2016-11-25 11:26:49

标签： scala apache-spark apache-spark-sql

民间，

我正在尝试在每行一个密钥标识符的spark中加入2个大型数据帧（每个100GB +）时遇到此问题。

我在EMR上使用Spark 1.6，这就是我正在做的事情：

val df1 = sqlContext.read.json("hdfs:///df1/")
val df2 = sqlContext.read.json("hdfs:///df2/")

// clean up and filter steps later 

df1.registerTempTable("df1")
df2.registerTempTable("df2")

val df3 = sql("select df1.*, df2.col1 from df1 left join df2 on df1.col3 = df2.col4")

df3.write.json("hdfs:///df3/")

这基本上是我正在做的事情的要点，以及最终加入df1和df2之间的其他清理和过滤步骤。

我看到的错误是：

java.lang.IllegalArgumentException: Size exceeds Integer.MAX_VALUE
    at sun.nio.ch.FileChannelImpl.map(FileChannelImpl.java:860)
    at org.apache.spark.storage.DiskStore$$anonfun$getBytes$2.apply(DiskStore.scala:127)
    at org.apache.spark.storage.DiskStore$$anonfun$getBytes$2.apply(DiskStore.scala:115)
    at org.apache.spark.util.Utils$.tryWithSafeFinally(Utils.scala:1250)
    at org.apache.spark.storage.DiskStore.getBytes(DiskStore.scala:129)
    at org.apache.spark.storage.DiskStore.getBytes(DiskStore.scala:136)
    at org.apache.spark.storage.BlockManager.doGetLocal(BlockManager.scala:503)
    at org.apache.spark.storage.BlockManager.getLocal(BlockManager.scala:420)
    at org.apache.spark.storage.BlockManager.get(BlockManager.scala:625)
    at org.apache.spark.CacheManager.getOrCompute(CacheManager.scala:44)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:268)
    at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
    at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
    at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
    at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
    at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
    at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
    at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:73)
    at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:41)
    at org.apache.spark.scheduler.Task.run(Task.scala:89)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:213)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
    at java.lang.Thread.run(Thread.java:745)

配置和参考：
我使用13个节点60GB每个群集与执行程序和驱动程序内存相应的开销设置。我试过调整的事情：

  • spark.sql.broadcastTimeout
  • spark.sql.shuffle.partitions

我也试过使用更大的集群，但没有帮助。 This link表示如果Shuffle分区大小超过2GB，则抛出此错误。但我已经尝试增加分区数量到一个很高的价值，但仍然没有运气。

我怀疑这可能与延迟加载有关。当我在DF上执行10次操作时，它们仅在最后一步执行。我尝试在DF的各种存储级别上添加.persist()，但它仍然没有成功。我也尝试删除临时表，清空所有早期的DF以进行清理。

然而，如果我将代码分为两部分 - 将最终的临时数据（2个数据帧）写入磁盘，则代码可以正常工作。重新启动只加入两个DF。

我之前收到此错误：

Exception in thread "main" java.util.concurrent.TimeoutException: Futures timed out after [300 seconds]
    at scala.concurrent.impl.Promise$DefaultPromise.ready(Promise.scala:219)
    at scala.concurrent.impl.Promise$DefaultPromise.result(Promise.scala:223)
    at scala.concurrent.Await$$anonfun$result$1.apply(package.scala:107)
    at scala.concurrent.BlockContext$DefaultBlockContext$.blockOn(BlockContext.scala:53)
    at scala.concurrent.Await$.result(package.scala:107)
    at org.apache.spark.sql.execution.joins.BroadcastHashOuterJoin.doExecute(BroadcastHashOuterJoin.scala:113)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$execute$5.apply(SparkPlan.scala:132)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$execute$5.apply(SparkPlan.scala:130)
    at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:150)
    at org.apache.spark.sql.execution.SparkPlan.execute(SparkPlan.scala:130)
    at org.apache.spark.sql.execution.Project.doExecute(basicOperators.scala:46)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$execute$5.apply(SparkPlan.scala:132)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$execute$5.apply(SparkPlan.scala:130)
    at org.apache.spark.rdd.RDDOperationScope$.withScope(RDDOperationScope.scala:150)
    at org.apache.spark.sql.execution.SparkPlan.execute(SparkPlan.scala:130)
    at org.apache.spark.sql.execution.QueryExecution.toRdd$lzycompute(QueryExecution.scala:55)
    at org.apache.spark.sql.execution.QueryExecution.toRdd(QueryExecution.scala:55)
    at org.apache.spark.sql.DataFrame.toJSON(DataFrame.scala:1724)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:606)
    at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:731)
    at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:181)
    at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:206)
    at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:121)
    at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)

但是当我调整spark.sql.broadcastTimeout时，我开始收到第一个错误。

在这种情况下会感谢任何帮助。如果需要，我可以添加更多信息。

1 个答案:

答案 0 :(得分：6)

在火花中你不能有大于2GB的随机阻挡。这是因为， Spark将shuffle块存储为ByteBuffer。这是你如何分配它：

ByteBuffer.allocate(int capacity)

因为，ByteBuffer受限于Integer.MAX_SIZE（2GB），所以是shuffle块！解决方案是通过在SparkSQL中使用spark.sql.shuffle.partitions或使用rdd.partition() or rdd.colease()为rdd来增加分区数，使每个分区大小<= 2GB。

您提到过您尝试增加分区数量，但仍然失败了。你能检查一下分区大小是否> 2GB。只需确保指定的分区数足以使每个块大小<1。 2GB

相关问题
最新问题