hezuojiao commented on a change in pull request #31898:
URL: https://github.com/apache/spark/pull/31898#discussion_r598297612
##########
File path:
core/src/main/scala/org/apache/spark/shuffle/BlockStoreShuffleReader.scala
##########
@@ -51,15 +51,17 @@ private[spark] class BlockStoreShuffleReader[K, C](
true
}
val useOldFetchProtocol = conf.get(config.SHUFFLE_USE_OLD_FETCH_PROTOCOL)
+ // SPARK-34790: Fetching continuous blocks in batch is incompatible with
io encryption.
+ val ioEncryption = conf.get(config.IO_ENCRYPTION_ENABLED)
Review comment:
Yes, you are right. This PR makes a simple read control, which may
sacrifice shuffle read performance in AQE. It would be better to allow batch
fetching for contiguous shuffle blocks while IO encryption enabled.
--
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]
