otterc commented on a change in pull request #35085:
URL: https://github.com/apache/spark/pull/35085#discussion_r799863347
##########
File path: core/src/main/scala/org/apache/spark/storage/DiskBlockManager.scala
##########
@@ -94,7 +95,13 @@ private[spark] class DiskBlockManager(
} else {
val newDir = new File(localDirs(dirId), "%02x".format(subDirId))
if (!newDir.exists()) {
- Files.createDirectory(newDir.toPath)
+ // SPARK-37618: Create dir as group writable so files within can be
deleted by the
+ // shuffle service
+ val path = newDir.toPath
+ Files.createDirectory(path)
+ val currentPerms = Files.getPosixFilePermissions(path)
+ currentPerms.add(PosixFilePermission.GROUP_WRITE)
Review comment:
@tgravescs Does Tez not rely on Yarn to remove the application local
dirs once the application finishes? If the TEZ DAG corresponds to a Yarn
application and if the shuffle data is written under the app local dirs, then
probably it is just leveraging Yarn to remove the entire app local dirs.
NodeManager deletes the app local dirs as the user using [this
code](https://github.com/apache/hadoop/blob/7c97c0f969e6e04513e0b5920855e3b7095e8d2b/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/native/container-executor/impl/container-executor.c#L2767),
I think. Even for Spark, this data gets deleted when the Spark application
finishes but in this change we are trying to remove the data earlier by the
Spark shuffle service.
--
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]
