[SPARK-34790] Fail in fetch shuffle blocks in batch when i/o encryption is enabled.

• Type: Sub-task
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 3.1.0, 3.1.1
• Fix Version/s: 3.1.2, 3.2.0
• Component/s: Spark Core
• Assignee: hezuojiao
• Reporter: hezuojiao
• Created: 18/Mar/21 16:07
• Updated: 22/Mar/21 20:36
• Resolved: 22/Mar/21 20:07

Description

When set spark.io.encryption.enabled=true, lots of test cases in AdaptiveQueryExecSuite will be failed. Fetching shuffle blocks in batch is incompatible with io encryption.

For example:
After set spark.io.encryption.enabled=true, run the following test suite which in AdaptiveQueryExecSuite:

 

  test("SPARK-33494: Do not use local shuffle reader for repartition") {
    withSQLConf(SQLConf.ADAPTIVE_EXECUTION_ENABLED.key -> "true") {
      val df = spark.table("testData").repartition('key)
      df.collect()
      // local shuffle reader breaks partitioning and shouldn't be used for repartition operation
      // which is specified by users.
      checkNumLocalShuffleReaders(df.queryExecution.executedPlan, numShufflesWithoutLocalReader = 1)
    }
  }

 
I got the following error message：

14:05:52.638 WARN org.apache.spark.scheduler.TaskSetManager: Lost task 1.0 in stage 2.0 (TID 3) (11.240.37.88 executor driver): FetchFailed(BlockManagerId(driver, 11.240.37.88, 63574, None), shuffleId=0, mapIndex=0, mapId=0, reduceId=2, message=14:05:52.638 WARN org.apache.spark.scheduler.TaskSetManager: Lost task 1.0 in stage 2.0 (TID 3) (11.240.37.88 executor driver): FetchFailed(BlockManagerId(driver, 11.240.37.88, 63574, None), shuffleId=0, mapIndex=0, mapId=0, reduceId=2, message=org.apache.spark.shuffle.FetchFailedException: Stream is corrupted at org.apache.spark.storage.ShuffleBlockFetcherIterator.throwFetchFailedException(ShuffleBlockFetcherIterator.scala:772) at org.apache.spark.storage.BufferReleasingInputStream.read(ShuffleBlockFetcherIterator.scala:845) at java.io.BufferedInputStream.fill(BufferedInputStream.java:246) at java.io.BufferedInputStream.read(BufferedInputStream.java:265) at java.io.DataInputStream.readInt(DataInputStream.java:387) at org.apache.spark.sql.execution.UnsafeRowSerializerInstance$$anon$2$$anon$3.readSize(UnsafeRowSerializer.scala:113) at org.apache.spark.sql.execution.UnsafeRowSerializerInstance$$anon$2$$anon$3.next(UnsafeRowSerializer.scala:129) at org.apache.spark.sql.execution.UnsafeRowSerializerInstance$$anon$2$$anon$3.next(UnsafeRowSerializer.scala:110) at scala.collection.Iterator$$anon$11.next(Iterator.scala:494) at scala.collection.Iterator$$anon$10.next(Iterator.scala:459) at org.apache.spark.util.CompletionIterator.next(CompletionIterator.scala:29) at org.apache.spark.InterruptibleIterator.next(InterruptibleIterator.scala:40) at scala.collection.Iterator$$anon$10.next(Iterator.scala:459) at org.apache.spark.sql.execution.SparkPlan.$anonfun$getByteArrayRdd$1(SparkPlan.scala:345) at org.apache.spark.rdd.RDD.$anonfun$mapPartitionsInternal$2(RDD.scala:898) at org.apache.spark.rdd.RDD.$anonfun$mapPartitionsInternal$2$adapted(RDD.scala:898) at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:52) at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:373) at org.apache.spark.rdd.RDD.iterator(RDD.scala:337) at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:90) at org.apache.spark.scheduler.Task.run(Task.scala:131) at org.apache.spark.executor.Executor$TaskRunner.$anonfun$run$3(Executor.scala:498) at org.apache.spark.util.Utils$.tryWithSafeFinally(Utils.scala:1437) at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:501) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748)Caused by: java.io.IOException: Stream is corrupted at net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:200) at net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:226) at net.jpountz.lz4.LZ4BlockInputStream.read(LZ4BlockInputStream.java:157) at org.apache.spark.storage.BufferReleasingInputStream.read(ShuffleBlockFetcherIterator.scala:841) ... 25 more
)

 

 

Issue Links

• is related to: SPARK-34827 Support fetching shuffle blocks in batch with i/o encryption
• links to: [Github] Pull Request #31898 (hezuojiao)
• links to: [Github] Pull Request #31898 (hezuojiao)