Github user zsxwing commented on a diff in the pull request: https://github.com/apache/spark/pull/20119#discussion_r159370490 --- Diff: core/src/main/scala/org/apache/spark/shuffle/IndexShuffleBlockResolver.scala --- @@ -198,7 +196,7 @@ private[spark] class IndexShuffleBlockResolver( // find out the consolidated file, then the offset within that from our index val indexFile = getIndexFile(blockId.shuffleId, blockId.mapId) - val in = new DataInputStream(Files.newInputStream(indexFile.toPath)) + val in = new DataInputStream(new FileInputStream(indexFile)) --- End diff -- @jerryshao this is another place. In addition, I'm not sure if there is any compression codec using `skip` or not. I also noticed `sun.nio.ch.ChannelInputStream` has extra `synchronized`s as `Files.newInputStream` needs to be thread-safe. Not sure if it may cause performance regression or not.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org