[#2510] improvement(common): group nearby local-order blocks into fewer RPCs - #2785
Open
sliortega295-ops wants to merge 1 commit into
Open
[#2510] improvement(common): group nearby local-order blocks into fewer RPCs#2785sliortega295-ops wants to merge 1 commit into
sliortega295-ops wants to merge 1 commit into
Conversation
…to fewer RPCs Generated-by: OpenAI Codex (GPT-5)
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
Allow the local-order segment splitter to keep small filtered gaps inside one contiguous shuffle-data read. Buffer-segment offsets now describe the expected blocks' positions inside that complete span, while storage boundaries and the configured read-buffer limit still split requests.
The fixed-size splitter keeps its existing contiguous-block behavior. Tests cover grouped gaps, exact and over-limit boundaries, trailing filtered blocks, storage changes, and the neighboring fixed-size path.
Why are the changes needed?
The previous implementation ended a segment at every filtered block. Workloads that select interleaved task attempts therefore issued many small local-data RPCs even when one bounded contiguous read could cover the same expected blocks. Grouping nearby blocks reduces request count while preserving the block metadata needed to ignore gap bytes.
Fixes #2510.
Does this PR introduce any user-facing change?
No public API or configuration change. Local-order shuffle reads may issue fewer, larger bounded requests.
How was this patch tested?
LocalOrderSegmentSplitterTest: 14 passedLocalOrderSegmentSplitterTestandFixedSizeSegmentSplitterTest: 19 passedspotless:checkandcheckstyle:check: passed with zero violationsgit diff --check: passedThe focused CPU tests demonstrate that expected blocks separated by small filtered gaps are served by one bounded read instead of separate requests. No GPU was used or required.
AI disclosure
The implementation, tests, and this description were generated with OpenAI Codex at the account owner's request. Codex re-checked the final diff against repository code and ran the validations listed above. No separate human line-by-line code review was performed before submission.