Ngone51 commented on a change in pull request #31763:
URL: https://github.com/apache/spark/pull/31763#discussion_r589899892
##########
File path: core/src/main/scala/org/apache/spark/scheduler/MapStatus.scala
##########
@@ -52,6 +52,13 @@ private[spark] sealed trait MapStatus {
* partitionId of the task or taskContext.taskAttemptId is used.
*/
def mapId: Long
+
+ /**
+ * Extra metadata for map status. This could be used by different
ShuffleManager implementation
+ * to store information they need. For example, a Remote Shuffle Service
ShuffleManager could
+ * store shuffle server information and let reducer task know where to fetch
shuffle data.
+ */
+ def metadata: Option[Serializable]
Review comment:
Then, it'd be a totally different topic, right? IIUC, SPARK-25299 could
also benefit custom shuffle manager if SPARK-25299(custom storage) is pluggable
with a custom shuffle manager. Ideally, a custom shuffle manager should be able
to plugin in different storages. We might not think deeply about how to support
custom shuffle managers when working on SPARK-25299 but need to keep in mind
the "pluggable". After SPARK-25299 completed, then, we can start to enhance the
support for custom shuffle managers.
That being said, I think you are still free to raise a separate discussion
on supporting the custom shuffle manager. For example, what are the
shortcomings of the current framework, and what should be improved...I just
wonder the community may not have enough bandwidth to work on these two
significant projects concurently.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
