[SPARK-40872][3.3] Fallback to original shuffle block when a push-merged shuffle chunk is zero-size#38751
Closed
gaoyajun02 wants to merge 1 commit intoapache:branch-3.3from
Closed
[SPARK-40872][3.3] Fallback to original shuffle block when a push-merged shuffle chunk is zero-size#38751gaoyajun02 wants to merge 1 commit intoapache:branch-3.3from
gaoyajun02 wants to merge 1 commit intoapache:branch-3.3from
Conversation
…huffle chunk is zero-size When push-based shuffle is enabled, a zero-size buf error may occur when fetching shuffle chunks from bad nodes, especially when memory is full. In this case, we can fall back to original shuffle blocks. When the reduce task obtains the shuffle chunk with a zero-size buf, we let it fall back to original shuffle block. After verification, these blocks can be read successfully without shuffle retry. No. UT Closes apache#38333 from gaoyajun02/SPARK-40872. Authored-by: gaoyajun02 <gaoyajun02@meituan.com> Signed-off-by: Mridul <mridul<at>gmail.com> (cherry picked from commit 72cce5c)
mridulm
approved these changes
Nov 22, 2022
Contributor
mridulm
left a comment
mridulm
left a comment
There was a problem hiding this comment.
Looks good to me.
Will wait for the tests to pass
Member
|
Seems like the test failure looks unrelated. I don't mind merging it as is. Feel free to retrigger https://github.com/gaoyajun02/spark/runs/9633118279 @gaoyajun02 |
|
Can one of the admins verify this patch? |
Contributor
Author
tests all passed @HyukjinKwon |
Member
|
Merged to branch-3.3. |
HyukjinKwon
pushed a commit
that referenced
this pull request
Nov 27, 2022
…ged shuffle chunk is zero-size ### What changes were proposed in this pull request? This is a backport PR of #38333. When push-based shuffle is enabled, a zero-size buf error may occur when fetching shuffle chunks from bad nodes, especially when memory is full. In this case, we can fall back to original shuffle blocks. ### Why are the changes needed? When the reduce task obtains the shuffle chunk with a zero-size buf, we let it fall back to original shuffle block. After verification, these blocks can be read successfully without shuffle retry. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? UT Closes #38751 from gaoyajun02/SPARK-40872-backport. Authored-by: gaoyajun02 <gaoyajun02@meituan.com> Signed-off-by: Hyukjin Kwon <gurwls223@apache.org>
Contributor
|
Thanks for the PR @gaoyajun02, and thanks for merging it @HyukjinKwon ! |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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 is a backport PR of #38333.
When push-based shuffle is enabled, a zero-size buf error may occur when fetching shuffle chunks from bad nodes, especially when memory is full. In this case, we can fall back to original shuffle blocks.
Why are the changes needed?
When the reduce task obtains the shuffle chunk with a zero-size buf, we let it fall back to original shuffle block. After verification, these blocks can be read successfully without shuffle retry.
Does this PR introduce any user-facing change?
No.
How was this patch tested?
UT