[Spark] Prevent FetchFailedException when enabling optimize write feature when running remote shuffle service (e.g. celeborn) #5681
+134
−34
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
zemin-piao
changed the title
Signed-off-by: zemin-piao <[email protected]>
…ss bins When using remote shuffle services (useShuffleManager=true), the ShuffleManager.getReader() API reads entire partitions. If bin-packing splits a reducer's blocks across multiple bins, each bin reads the full partition, causing data duplication in output files. This fix changes bin-packing behavior for remote shuffle mode: - Large reducers (>= binSize): each gets its own bin - Small reducers (< binSize): bin-packed together as atomic units - Reducers are never split across bins Trade-off: Large reducers may produce oversized files, but this is acceptable compared to data duplication. Local shuffle mode (default) is unchanged and continues to bin-pack individual blocks for optimal file sizes.
zemin-piao
force-pushed
the
optimize_write_for_remote_shuffle_service
branch
from
eb9cf29 to
2f4b8ea
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Which Delta project/connector is this regarding?
Description
Resolves #4343
How was this patch tested?
This was tested with running apache spark + deltalake + celeborn with client mode. No row duplicates + no shuffle fetch failures happens
Does this PR introduce any user-facing changes?
Yes, previous behavior is that when the optimizeWrite feature is enabled + when running in parallel remote shuffle service for spark, the
FetchFailedExceptionhappens. After this PR, by specifying the flaguseShuffleManagerto betrue, theFetchFailedExceptionis prevented even using remote shuffle service (e.g. apache celeborn)