Github user mridulm commented on a diff in the pull request:
https://github.com/apache/spark/pull/16989#discussion_r115826803
--- Diff:
core/src/main/scala/org/apache/spark/storage/ShuffleBlockFetcherIterator.scala
---
@@ -175,33 +193,54 @@ final class ShuffleBlockFetcherIterator(
val sizeMap = req.blocks.map { case (blockId, size) =>
(blockId.toString, size) }.toMap
val remainingBlocks = new HashSet[String]() ++= sizeMap.keys
val blockIds = req.blocks.map(_._1.toString)
-
val address = req.address
- shuffleClient.fetchBlocks(address.host, address.port,
address.executorId, blockIds.toArray,
- 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 {
- 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)
- }
+
+ 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 {
+ 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)
}
- logTrace("Got remote block " + blockId + " after " +
Utils.getUsedTimeMs(startTime))
}
+ 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))
- }
+ 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))
}
- )
+ }
+
+ // Shuffle remote blocks to disk when the request is too large or
local memory shortage.
+ val fetchToDisk = if (bytesInFlight > maxBytesInFlight) {
--- End diff --
`maxBytesInFlight` is supposed to control how much data we want in flight
(on network) - and not how much data should be used for shuffle in memory.
Prior to this change, there was a direct correlation due to the
implementation details (you can only shuffle what you have read).
Though related, they might need to be independently controlled - or are we
tying them together for simplicity for now ? Any thoughts @cloud-fan ?
For example: default value of maxBytesInFlight is 48mb iirc - which is can
be really small for executors running on 10s or 100s of GB RAM !
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
