sarutak opened a new pull request, #53940:
URL: https://github.com/apache/spark/pull/53940
### What changes were proposed in this pull request?
This PR fixes an issue that `unlock` and `releaseAllLocksForTask` for write
lock in `BlockInfoManager`.
#52524 tried to fix this issue but it's still remaining.
This issue can be reproduced by inserting `Thread.sleep()` into
`BlockInfoManager` like as follows.
```
--- a/core/src/main/scala/org/apache/spark/storage/BlockInfoManager.scala
+++ b/core/src/main/scala/org/apache/spark/storage/BlockInfoManager.scala
@@ -397,6 +397,7 @@ private[storage] class
BlockInfoManager(trackingCacheVisibility: Boolean = false
blockInfo(blockId) { (info, condition) =>
if (info.writerTask != BlockInfo.NO_WRITER) {
info.writerTask = BlockInfo.NO_WRITER
+ Thread.sleep(100)
val blockIds = writeLocksByTask.get(taskAttemptId)
if (blockIds != null) {
blockIds.remove(blockId)
@@ -490,12 +491,14 @@ private[storage] class
BlockInfoManager(trackingCacheVisibility: Boolean = false
val blocksWithReleasedLocks = mutable.ArrayBuffer[BlockId]()
val writeLocks =
Option(writeLocksByTask.remove(taskAttemptId)).getOrElse(util.Set.of())
+ Thread.sleep(100)
writeLocks.forEach { blockId =>
blockInfo(blockId) { (info, condition) =>
// Check the existence of `blockId` because `unlock` may have
already removed it
// concurrently.
if (writeLocks.contains(blockId)) {
blocksWithReleasedLocks += blockId
+ Thread.sleep(100)
assert(info.writerTask == taskAttemptId)
info.writerTask = BlockInfo.NO_WRITER
condition.signalAll()
```
And then, run a test `SPARK-53807 - concurrent unlock and
releaseAllLocksForTask for write should not fail` in `BlockInfoManagerSuite`.
```
$ build/sbt 'testOnly org.apache.spark.storage.BlockInfoManagerSuite -- -z
"SPARK-53807"'
```
### Why are the changes needed?
To fix a race condition issue.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
`BlockInfoManagerSuite` passes even if inserting `Thread.sleep` as mentioned
above.
### Was this patch authored or co-authored using generative AI tooling?
No.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
