suneet-s commented on a change in pull request #11492:
URL: https://github.com/apache/druid/pull/11492#discussion_r676861665
##########
File path:
indexing-service/src/main/java/org/apache/druid/indexing/worker/shuffle/ShuffleResource.java
##########
@@ -98,14 +96,19 @@ public Response getPartition(
);
return Response.status(Status.NOT_FOUND).entity(errorMessage).build();
} else {
- shuffleMetrics.ifPresent(metrics ->
metrics.shuffleRequested(supervisorTaskId, partitionFile.length()));
- return Response.ok(
- (StreamingOutput) output -> {
- try (final FileInputStream fileInputStream = new
FileInputStream(partitionFile)) {
- ByteStreams.copy(fileInputStream, output);
- }
- }
- ).build();
+ try {
+ long size = partitionFile.get().size();
Review comment:
I was looking at the default implementation of `ByteSource#size()` which
opens the stream, reads it and counts the bytes. It looks like `FileByteSource`
has overridden that implementation to return the length of the File without
reading the bytes.
At this point, I'm mostly concerned about whether someone can trip over this
by changing the implementation and not realizing that this operation should be
fast. Anyways, that's a problem for another time.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
