cloud-fan commented on a change in pull request #24565: [SPARK-27665][Core] Split fetch shuffle blocks protocol from OpenBlocks URL: https://github.com/apache/spark/pull/24565#discussion_r283249401
########## File path: docs/sql-migration-guide-upgrade.md ########## @@ -128,6 +128,8 @@ license: | - Since Spark 3.0, if `hive.default.fileformat` is not found in `Spark SQL configuration` then it will fallback to hive-site.xml present in the `Hadoop configuration` of `SparkContext`. + - Since Spark 3.0, for the better extension for fetching shuffle block, we split the protocol into two parts. The protocol change is transparent for the users without external shuffle service, but for the others, we need to upgrade the external shuffle service jar when you see the error message like `java.lang.UnsupportedOperationException: Unexpected message: FetchShuffleBlocks`. If the Spark version is 2.4 or earlier, but external shuffle service is 3.0, we support this scenario by doing the compatible work for OpenBlocks inside `ExternalShuffleBlockHandler`. Review comment: The migration guide is for end users. Let's refine this and only include the information that for end-users, without talking too much about internals. ---------------------------------------------------------------- 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] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
