xkrogen commented on a change in pull request #32389:
URL: https://github.com/apache/spark/pull/32389#discussion_r633725651
##########
File path:
core/src/test/scala/org/apache/spark/storage/ShuffleBlockFetcherIteratorSuite.scala
##########
@@ -123,6 +131,42 @@ class ShuffleBlockFetcherIteratorSuite extends
SparkFunSuite with PrivateMethodT
verify(wrappedInputStream.invokePrivate(delegateAccess()),
times(1)).close()
}
+ // scalastyle:off argcount
+ private def createShuffleBlockIteratorWithDefaults(
+ blocksByAddress: Map[BlockManagerId, (Traversable[BlockId], Long, Int)],
+ taskContext: Option[TaskContext] = None,
+ streamWrapperLimitSize: Option[Long] = None,
+ blockManager: Option[BlockManager] = None,
+ maxBytesInFlight: Long = Long.MaxValue,
+ maxReqsInFlight: Int = Int.MaxValue,
+ maxBlocksInFlightPerAddress: Int = Int.MaxValue,
+ maxReqSizeShuffleToMem: Int = Int.MaxValue,
+ detectCorrupt: Boolean = true,
+ detectCorruptUseExtraMemory: Boolean = true,
+ shuffleMetrics: Option[ShuffleReadMetricsReporter] = None,
+ doBatchFetch: Boolean = false): ShuffleBlockFetcherIterator = {
+ val tContext = taskContext.getOrElse(TaskContext.empty())
+ new ShuffleBlockFetcherIterator(
+ tContext,
+ transfer,
+ blockManager.getOrElse(createMockBlockManager()),
+ blocksByAddress.map { case (blockManagerId, (blocks, blockSize,
blockMapIndex)) =>
+ (blockManagerId, blocks.map(blockId => (blockId, blockSize,
blockMapIndex)).toSeq)
+ }.toIterator,
+ streamWrapperLimitSize
+ .map(limit => (_: BlockId, in: InputStream) => new
LimitedInputStream(in, limit))
+ .getOrElse((_: BlockId, in: InputStream) => in),
Review comment:
good suggestion! incorporated.
##########
File path:
core/src/test/scala/org/apache/spark/storage/ShuffleBlockFetcherIteratorSuite.scala
##########
@@ -703,40 +600,24 @@ class ShuffleBlockFetcherIteratorSuite extends
SparkFunSuite with PrivateMethodT
val sem = new Semaphore(0)
val corruptLocalBuffer = new FileSegmentManagedBuffer(null, new File("a"),
0, 100)
- val transfer = mock(classOf[BlockTransferService])
- when(transfer.fetchBlocks(any(), any(), any(), any(), any(), any()))
- .thenAnswer((invocation: InvocationOnMock) => {
- val listener =
invocation.getArguments()(4).asInstanceOf[BlockFetchingListener]
- Future {
- // Return the first block, and then fail.
- listener.onBlockFetchSuccess(
- ShuffleBlockId(0, 0, 0).toString, blocks(ShuffleBlockId(0, 0, 0)))
- listener.onBlockFetchSuccess(
- ShuffleBlockId(0, 1, 0).toString, mockCorruptBuffer())
- listener.onBlockFetchSuccess(
- ShuffleBlockId(0, 2, 0).toString, corruptLocalBuffer)
- sem.release()
- }
- })
-
- val blocksByAddress = Seq[(BlockManagerId, Seq[(BlockId, Long, Int)])](
- (remoteBmId, blocks.keys.map(blockId => (blockId, 1L,
0)).toSeq)).toIterator
+ answerFetchBlocks { invocation =>
+ val listener = invocation.getArgument[BlockFetchingListener](4)
+ Future {
+ // Return the first block, and then fail.
+ listener.onBlockFetchSuccess(
+ ShuffleBlockId(0, 0, 0).toString, blocks(ShuffleBlockId(0, 0, 0)))
+ listener.onBlockFetchSuccess(
+ ShuffleBlockId(0, 1, 0).toString, mockCorruptBuffer())
+ listener.onBlockFetchSuccess(
+ ShuffleBlockId(0, 2, 0).toString, corruptLocalBuffer)
+ sem.release()
+ }
+ }
- val taskContext = TaskContext.empty()
- val iterator = new ShuffleBlockFetcherIterator(
- taskContext,
- transfer,
- blockManager,
- blocksByAddress,
- (_, in) => new LimitedInputStream(in, 100),
- 48 * 1024 * 1024,
- Int.MaxValue,
- Int.MaxValue,
- Int.MaxValue,
- true,
- true,
- taskContext.taskMetrics.createTempShuffleReadMetrics(),
- false)
+ val iterator = createShuffleBlockIteratorWithDefaults(
+ Map(remoteBmId ->(blocks.keys, 1L, 0)),
Review comment:
thanks!
##########
File path:
core/src/test/scala/org/apache/spark/storage/ShuffleBlockFetcherIteratorSuite.scala
##########
@@ -123,6 +131,42 @@ class ShuffleBlockFetcherIteratorSuite extends
SparkFunSuite with PrivateMethodT
verify(wrappedInputStream.invokePrivate(delegateAccess()),
times(1)).close()
}
+ // scalastyle:off argcount
+ private def createShuffleBlockIteratorWithDefaults(
+ blocksByAddress: Map[BlockManagerId, (Traversable[BlockId], Long, Int)],
+ taskContext: Option[TaskContext] = None,
+ streamWrapperLimitSize: Option[Long] = None,
+ blockManager: Option[BlockManager] = None,
+ maxBytesInFlight: Long = Long.MaxValue,
+ maxReqsInFlight: Int = Int.MaxValue,
+ maxBlocksInFlightPerAddress: Int = Int.MaxValue,
+ maxReqSizeShuffleToMem: Int = Int.MaxValue,
+ detectCorrupt: Boolean = true,
+ detectCorruptUseExtraMemory: Boolean = true,
+ shuffleMetrics: Option[ShuffleReadMetricsReporter] = None,
+ doBatchFetch: Boolean = false): ShuffleBlockFetcherIterator = {
+ val tContext = taskContext.getOrElse(TaskContext.empty())
+ new ShuffleBlockFetcherIterator(
+ tContext,
+ transfer,
+ blockManager.getOrElse(createMockBlockManager()),
+ blocksByAddress.map { case (blockManagerId, (blocks, blockSize,
blockMapIndex)) =>
+ (blockManagerId, blocks.map(blockId => (blockId, blockSize,
blockMapIndex)).toSeq)
+ }.toIterator,
Review comment:
I think this is closely related to your [other
comment](https://github.com/apache/spark/pull/32389#discussion_r632823782), so
I'll respond to both here.
I agree that this is tailored to the current set of tests and thus current
usage, as opposed to potential future usage. If we were designing a public API,
or even a private API in the production code, I would agree with you. But in
this case for a class-private method in a test file, I'm not convinced that
designing for future possibilities is the right move. If someone later adds
tests which do need blocks of different sizes, they can make the modification
you've described, right? For now I would preference simplicity, and we can
introduce the additional complexity if necessary.
WDYT? This is not a strong conviction on my side so if you are not convinced
I can make changes as you proposed.
##########
File path:
core/src/test/scala/org/apache/spark/storage/ShuffleBlockFetcherIteratorSuite.scala
##########
@@ -123,6 +131,42 @@ class ShuffleBlockFetcherIteratorSuite extends
SparkFunSuite with PrivateMethodT
verify(wrappedInputStream.invokePrivate(delegateAccess()),
times(1)).close()
}
+ // scalastyle:off argcount
+ private def createShuffleBlockIteratorWithDefaults(
+ blocksByAddress: Map[BlockManagerId, (Traversable[BlockId], Long, Int)],
+ taskContext: Option[TaskContext] = None,
+ streamWrapperLimitSize: Option[Long] = None,
+ blockManager: Option[BlockManager] = None,
+ maxBytesInFlight: Long = Long.MaxValue,
+ maxReqsInFlight: Int = Int.MaxValue,
+ maxBlocksInFlightPerAddress: Int = Int.MaxValue,
+ maxReqSizeShuffleToMem: Int = Int.MaxValue,
+ detectCorrupt: Boolean = true,
+ detectCorruptUseExtraMemory: Boolean = true,
+ shuffleMetrics: Option[ShuffleReadMetricsReporter] = None,
+ doBatchFetch: Boolean = false): ShuffleBlockFetcherIterator = {
+ val tContext = taskContext.getOrElse(TaskContext.empty())
+ new ShuffleBlockFetcherIterator(
+ tContext,
+ transfer,
+ blockManager.getOrElse(createMockBlockManager()),
+ blocksByAddress.map { case (blockManagerId, (blocks, blockSize,
blockMapIndex)) =>
+ (blockManagerId, blocks.map(blockId => (blockId, blockSize,
blockMapIndex)).toSeq)
+ }.toIterator,
+ streamWrapperLimitSize
+ .map(limit => (_: BlockId, in: InputStream) => new
LimitedInputStream(in, limit))
+ .getOrElse((_: BlockId, in: InputStream) => in),
+ maxBytesInFlight,
+ maxReqsInFlight,
+ maxBlocksInFlightPerAddress,
+ maxReqSizeShuffleToMem,
+ detectCorrupt,
+ detectCorruptUseExtraMemory,
+
shuffleMetrics.getOrElse(tContext.taskMetrics().createTempShuffleReadMetrics()),
Review comment:
`taskMetrics()` is an empty-paren method so it should be called with
parenthesis:
```
def taskMetrics(): TaskMetrics
```
##########
File path:
core/src/test/scala/org/apache/spark/storage/ShuffleBlockFetcherIteratorSuite.scala
##########
@@ -123,6 +131,42 @@ class ShuffleBlockFetcherIteratorSuite extends
SparkFunSuite with PrivateMethodT
verify(wrappedInputStream.invokePrivate(delegateAccess()),
times(1)).close()
}
+ // scalastyle:off argcount
+ private def createShuffleBlockIteratorWithDefaults(
+ blocksByAddress: Map[BlockManagerId, (Traversable[BlockId], Long, Int)],
+ taskContext: Option[TaskContext] = None,
+ streamWrapperLimitSize: Option[Long] = None,
+ blockManager: Option[BlockManager] = None,
+ maxBytesInFlight: Long = Long.MaxValue,
+ maxReqsInFlight: Int = Int.MaxValue,
+ maxBlocksInFlightPerAddress: Int = Int.MaxValue,
+ maxReqSizeShuffleToMem: Int = Int.MaxValue,
+ detectCorrupt: Boolean = true,
+ detectCorruptUseExtraMemory: Boolean = true,
+ shuffleMetrics: Option[ShuffleReadMetricsReporter] = None,
+ doBatchFetch: Boolean = false): ShuffleBlockFetcherIterator = {
+ val tContext = taskContext.getOrElse(TaskContext.empty())
+ new ShuffleBlockFetcherIterator(
+ tContext,
+ transfer,
+ blockManager.getOrElse(createMockBlockManager()),
+ blocksByAddress.map { case (blockManagerId, (blocks, blockSize,
blockMapIndex)) =>
+ (blockManagerId, blocks.map(blockId => (blockId, blockSize,
blockMapIndex)).toSeq)
+ }.toIterator,
+ streamWrapperLimitSize
+ .map(limit => (_: BlockId, in: InputStream) => new
LimitedInputStream(in, limit))
+ .getOrElse((_: BlockId, in: InputStream) => in),
+ maxBytesInFlight,
+ maxReqsInFlight,
+ maxBlocksInFlightPerAddress,
+ maxReqSizeShuffleToMem,
+ detectCorrupt,
+ detectCorruptUseExtraMemory,
+
shuffleMetrics.getOrElse(tContext.taskMetrics().createTempShuffleReadMetrics()),
Review comment:
`taskMetrics()` is defined as an empty-paren method so it should be
called with parenthesis:
```
def taskMetrics(): TaskMetrics
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
