cloud-fan commented on a change in pull request #24892: [SPARK-25341][Core]
Support rolling back a shuffle map stage and re-generate the shuffle files
URL: https://github.com/apache/spark/pull/24892#discussion_r300279227
##########
File path:
common/network-shuffle/src/main/java/org/apache/spark/network/shuffle/ShuffleClient.java
##########
@@ -43,6 +44,18 @@
* they will be kept in memory.
*/
public abstract void fetchBlocks(
+ String host,
+ int port,
+ String execId,
+ int shuffleGenerationId,
+ String[] blockIds,
+ BlockFetchingListener listener,
+ DownloadFileManager downloadFileManager);
+
+ /**
+ * Fetch a sequence of non-shuffle blocks from a remote node asynchronously.
+ */
+ public abstract void fetchDataBlocks(
Review comment:
This is a little weird, but according to SPARK-27677, this is by design: we
want external shuffle service to serve cached RDD blocks. Does normal shuffle
also use `ShuffleClient`?
----------------------------------------------------------------
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]
