[CELEBORN-2166] Fast fail reduce stage if shuffle data is lost because of worker lost #3496
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
…e of worker crash/lost
SteNicholas
reviewed
Oct 10, 2025
client/src/main/scala/org/apache/celeborn/client/commit/ReducePartitionCommitHandler.scala
Show resolved
Hide resolved
SteNicholas
reviewed
Oct 10, 2025
SteNicholas
changed the title
Member
|
Merged to main(v0.7.0). |
Member
|
We can merge this PR into branch-0.6 as well, will update the config version. |
turboFei
pushed a commit
that referenced
this pull request
Dec 30, 2025
…e of worker lost - Fix the WorkerStatusTracker logic, so unknown workers are marked correctly in excluded workers. - Trigger shuffle data lost if the worker hosting the shuffle data is lost. This can be extended to – - fast fail mapper stages as well before the commit starts. - with push replicate enabled with multiple workers loss. Currently even if worker crashs or became unavailable for some reason and marked as lost by Master, reduce stage still try to read data from it and fail after running for sometime which is in-efficient. We can detect this early and fail the reduce stage with SHUFFLE_DATA_LOST before starting the stage. NA WIP Closes #3496 from s0nskar/CELEBORN-2166. Authored-by: Sanskar Modi <sanskarmodi97@gmail.com> Signed-off-by: SteNicholas <programgeek@163.com> (cherry picked from commit 1157d6a) Signed-off-by: Wang, Fei <fwang12@ebay.com>
Member
|
thanks, merged to 0.6.3 as well |
turboFei
added a commit
that referenced
this pull request
Dec 30, 2025
…stOnUnknownWorker.enabled version to 0.6.3 ### What changes were proposed in this pull request? Update config celeborn.client.shuffleDataLostOnUnknownWorker.enabled version to 0.6.3 ### Why are the changes needed? Followup for #3496, it is better to merge into branch-0.6 as well. ### Does this PR resolve a correctness bug? No. ### Does this PR introduce _any_ user-facing change? No, it has not been releases yet. ### How was this patch tested? GA. Closes #3576 from turboFei/update_conf. Authored-by: Wang, Fei <fwang12@ebay.com> Signed-off-by: Wang, Fei <fwang12@ebay.com>
turboFei
added a commit
that referenced
this pull request
Dec 30, 2025
…stOnUnknownWorker.enabled version to 0.6.3 ### What changes were proposed in this pull request? Update config celeborn.client.shuffleDataLostOnUnknownWorker.enabled version to 0.6.3 ### Why are the changes needed? Followup for #3496, it is better to merge into branch-0.6 as well. ### Does this PR resolve a correctness bug? No. ### Does this PR introduce _any_ user-facing change? No, it has not been releases yet. ### How was this patch tested? GA. Closes #3576 from turboFei/update_conf. Authored-by: Wang, Fei <fwang12@ebay.com> Signed-off-by: Wang, Fei <fwang12@ebay.com> (cherry picked from commit 38532d7) Signed-off-by: Wang, Fei <fwang12@ebay.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 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.
What changes were proposed in this pull request?
This can be extended to –
Why are the changes needed?
Currently even if worker crashs or became unavailable for some reason and marked as lost by Master, reduce stage still try to read data from it and fail after running for sometime which is in-efficient. We can detect this early and fail the reduce stage with SHUFFLE_DATA_LOST before starting the stage.
Does this PR introduce any user-facing change?
NA
How was this patch tested?
WIP