Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/10705#discussion_r53729151
--- Diff: core/src/main/scala/org/apache/spark/CacheManager.scala ---
@@ -46,7 +48,14 @@ private[spark] class CacheManager(blockManager:
BlockManager) extends Logging {
val existingMetrics =
context.taskMetrics().registerInputMetrics(blockResult.readMethod)
existingMetrics.incBytesReadInternal(blockResult.bytes)
- val iter = blockResult.data.asInstanceOf[Iterator[T]]
+ val iter = {
+ val dataIter = blockResult.data.asInstanceOf[Iterator[T]]
+ if (blockResult.readMethod != DataReadMethod.Network) {
+ CompletionIterator[T, Iterator[T]](dataIter,
blockManager.releaseLock(key))
--- End diff --
In 66202f2d7eda4abcf1b315a85f91b43d4424d93e, I updated BlockManager to pass
CompletionIterators to BlockResult. This felt a bit cleaner than putting a
BlockManager reference into the BlockResult constructor in order to be able to
reference it in the completion iterator, although I suppose that I could call
SparkEnv.get.blockManager... from the completion iterator's cleanup task
instead.
---
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]
