[SPARK-25070] BlockFetchingListener#onBlockFetchSuccess throw "java.util.NoSuchElementException: key not found: shuffle_8_68_113" on ShuffleBlockFetcherIterator caused stage hang long time

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Incomplete
• Affects Version/s: 2.2.0
• Component/s: Spark Core
• Labels: bulk-closed
• Assignee: Unassigned
• Reporter: DENG FEI
• Created: 09/Aug/18 09:51
• Updated: 08/Oct/19 05:42
• Resolved: 08/Oct/19 05:42

Description

The task fetch shuffle block success, but failed onBlockFetchSuccess, lead the task hang long time and speculate as false.

The log is below:

18/08/08 14:55:53 INFO ShuffleBlockFetcherIterator: Started 16 remote fetches in 16 ms

18/08/08 14:55:53 WARN TransportChannelHandler: Exception in connection from /xxx.xxx.xxx.xxx:7337 java.util.NoSuchElementException: key not found: shuffle_8_68_113 at scala.collection.MapLike$class.default(MapLike.scala:228) at scala.collection.AbstractMap.default(Map.scala:59) at scala.collection.MapLike$class.apply(MapLike.scala:141) at scala.collection.AbstractMap.apply(Map.scala:59) at org.apache.spark.storage.ShuffleBlockFetcherIterator$$anon$1.onBlockFetchSuccess(ShuffleBlockFetcherIterator.scala:217) at org.apache.spark.network.shuffle.RetryingBlockFetcher$RetryingBlockFetchListener.onBlockFetchSuccess(RetryingBlockFetcher.java:204) at org.apache.spark.network.shuffle.OneForOneBlockFetcher$ChunkCallback.onSuccess(OneForOneBlockFetcher.java:97) at org.apache.spark.network.client.TransportResponseHandler.handle(TransportResponseHandler.java:171) at

XXXXXX

18/08/08 14:55:53 INFO Executor: Finished task 44.0 in stage 14.0 (TID 1483). 3458 bytes result sent to driver 18/08/09 10:02:32 INFO Executor: Executor is trying to kill task 113.0 in stage 14.0 (TID 1552), reason: stage cancelled

val blockFetchingListener = new BlockFetchingListener {
  override def onBlockFetchSuccess(blockId: String, buf: ManagedBuffer): Unit = {
    // Only add the buffer to results queue if the iterator is not zombie,
    // i.e. cleanup() has not been called yet.
    ShuffleBlockFetcherIterator.this.synchronized {
      try {
        if (!isZombie) {
          // Increment the ref count because we need to pass this to a different thread.
          // This needs to be released after use.
          buf.retain()
          remainingBlocks -= blockId
          results.put(new SuccessFetchResult(BlockId(blockId), address, sizeMap(blockId), buf,
            remainingBlocks.isEmpty))
          logDebug("remainingBlocks: " + remainingBlocks)
        }
      } catch {
        case e : Throwable => onBlockFetchFailure(blockId, e)
      }
    }
    logTrace("Got remote block " + blockId + " after " + Utils.getUsedTimeMs(startTime))
  }

  override def onBlockFetchFailure(blockId: String, e: Throwable): Unit = {
    logError(s"Failed to get block(s) from ${req.address.host}:${req.address.port}", e)
    results.put(new FailureFetchResult(BlockId(blockId), address, e))
  }
}

 

 

results.put(new SuccessFetchResult(BlockId(blockId), address, sizeMap(blockId), buf,
            remainingBlocks.isEmpty))