otterc commented on a change in pull request #32140:
URL: https://github.com/apache/spark/pull/32140#discussion_r645674391
##########
File path:
common/network-shuffle/src/main/java/org/apache/spark/network/shuffle/ExternalBlockHandler.java
##########
@@ -413,6 +466,47 @@ public ManagedBuffer next() {
}
}
+ private class ShuffleChunkManagedBufferIterator implements
Iterator<ManagedBuffer> {
+
+ private int reduceIdx = 0;
+ private int chunkIdx = 0;
+
+ private final String appId;
+ private final int shuffleId;
+ private final int[] reduceIds;
+ private final int[][] chunkIds;
+
+ ShuffleChunkManagedBufferIterator(FetchShuffleBlockChunks msg) {
+ appId = msg.appId;
+ shuffleId = msg.shuffleId;
+ reduceIds = msg.reduceIds;
+ chunkIds = msg.chunkIds;
+ }
+
+ @Override
+ public boolean hasNext() {
+ // reduceIds.length must equal to chunkIds.length, and the passed in
FetchShuffleBlockChunks
+ // must have non-empty reduceIds and chunkIds, see the checking logic in
+ // OneForOneBlockFetcher.
+ assert(reduceIds.length != 0 && reduceIds.length == chunkIds.length);
+ return reduceIdx < reduceIds.length && chunkIdx <
chunkIds[reduceIdx].length;
+ }
+
+ @Override
+ public ManagedBuffer next() {
+ ManagedBuffer block = mergeManager.getMergedBlockData(
+ appId, shuffleId, reduceIds[reduceIdx], chunkIds[reduceIdx][chunkIdx]);
+ if (chunkIdx < chunkIds[reduceIdx].length - 1) {
+ chunkIdx += 1;
+ } else {
+ chunkIdx = 0;
+ reduceIdx += 1;
+ }
+ metrics.blockTransferRateBytes.mark(block != null ? block.size() : 0);
Review comment:
It's not going to be null. The implementation either throws an exception
or returns a `FileSegmentManagedBuffer`. However, even the other iterators
`ManagerBufferIterator` and `ShuffleManagedBufferIterator` check if the block
is null even though `blockManager.getBlockData` and
`blockManager.getContinuousBlocksData` will not return null.
Please let me know if I should remove this check for
`ShuffleChunkManagedBufferIterator`?
--
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:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
