Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/9000#discussion_r41570238
--- Diff: core/src/main/scala/org/apache/spark/storage/MemoryStore.scala ---
@@ -525,22 +519,40 @@ private[spark] class MemoryStore(blockManager:
BlockManager, maxMemory: Long)
/**
* Reserve the unroll memory of current unroll successful block used by
this task
* until actually put the block into memory entry.
+ * @return whether the request is granted.
*/
- def reservePendingUnrollMemoryForThisTask(memory: Long): Unit = {
+ private def reservePendingUnrollMemoryForThisTask(
+ blockId: BlockId,
+ memory: Long,
+ droppedBlocks: mutable.Buffer[(BlockId, BlockStatus)]): Boolean = {
val taskAttemptId = currentTaskAttemptId()
accountingLock.synchronized {
- pendingUnrollMemoryMap(taskAttemptId) =
- pendingUnrollMemoryMap.getOrElse(taskAttemptId, 0L) + memory
+ val acquired = memoryManager.acquireUnrollMemory(blockId, memory,
droppedBlocks)
+ val success = acquired == memory
+ if (success) {
+ pendingUnrollMemoryMap(taskAttemptId) =
+ pendingUnrollMemoryMap.getOrElse(taskAttemptId, 0L) + memory
+ }
--- End diff --
I just noticed that the return value of
`reservePendingUnrollMemoryForThisTask` is never checked. Does this mean that
we never expect this operation to fail?
---
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]
