[fix](nereids) Refine aggregate shuffle-key reuse safety checks - #64892
Open
foxtail463 wants to merge 5 commits into
Open
[fix](nereids) Refine aggregate shuffle-key reuse safety checks#64892foxtail463 wants to merge 5 commits into
foxtail463 wants to merge 5 commits into
Conversation
foxtail463
requested review from
924060929,
englefly,
morrySnow and
starocean999
as code owners
June 26, 2026 09:15
Contributor
|
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 29420 ms |
Contributor
TPC-DS: Total hot run time: 171785 ms |
foxtail463
force-pushed
the
fix/cbo-agg-parent-shuffle-distinct-oom
branch
from
June 26, 2026 09:55
019cd1b to
2de423e
Compare
Contributor
ClickBench: Total hot run time: 25.24 s |
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 28727 ms |
Contributor
TPC-DS: Total hot run time: 171183 ms |
Contributor
ClickBench: Total hot run time: 25.15 s |
Contributor
FE Regression Coverage ReportIncrement line coverage |
morrySnow
reviewed
Jun 29, 2026
morrySnow
left a comment
morrySnow
left a comment
Contributor
There was a problem hiding this comment.
Thanks for this fix! The change from default-true to default-false in shouldUseParent when stats are missing is the right safety trade-off — preventing OOM is much more important than a potentially narrower shuffle. The approach of pre-resolving parent hash expressions against group-by expressions in visitPhysicalHashAggregate also simplifies shouldUseParent nicely.
I left a few inline comments for your consideration.
| physicalHashAggregate( | ||
| physicalDistribute(any())))), | ||
| physicalDistribute( | ||
| physicalHashAggregate( |
Contributor
There was a problem hiding this comment.
The physicalDistribute wrappers are now expected in the plan shape because shouldUseParent no longer returns true when stats are unknown (which is the case in this unit test). Previously, the parent subset key was blindly adopted, which could eliminate the distribute node. This test change correctly reflects the stricter stats gate — the full group-by key distribution is used, and the distribute is preserved.
This is an intended side effect of the fix, but worth confirming: is the plan shape here what you would expect to see in production queries after this change?
Contributor
Author
|
run buildall |
Contributor
FE UT Coverage ReportIncrement line coverage |
Contributor
TPC-H: Total hot run time: 29567 ms |
Contributor
TPC-DS: Total hot run time: 180211 ms |
Contributor
ClickBench: Total hot run time: 24.95 s |
Contributor
FE Regression Coverage ReportIncrement line coverage |
foxtail463
force-pushed
the
fix/cbo-agg-parent-shuffle-distinct-oom
branch
from
August 3, 2026 10:44
b41acce to
06375c3
Compare
Contributor
Author
|
run buildall |
morrySnow
changed the title
Contributor
TPC-H: Total hot run time: 29222 ms |
Contributor
TPC-DS: Total hot run time: 168818 ms |
Contributor
ClickBench: Total hot run time: 23.71 s |
Contributor
FE Regression Coverage ReportIncrement line coverage |
Contributor
TPC-H: Total hot run time: 16819 ms |
Contributor
TPC-DS: Total hot run time: 83970 ms |
Contributor
ClickBench: Total hot run time: 17.48 s |
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 17071 ms |
Contributor
TPC-DS: Total hot run time: 86176 ms |
Contributor
ClickBench: Total hot run time: 17.74 s |
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 16872 ms |
Contributor
TPC-DS: Total hot run time: 84387 ms |
Contributor
ClickBench: Total hot run time: 17.71 s |
Contributor
Author
|
run buildall |
foxtail463
force-pushed
the
fix/cbo-agg-parent-shuffle-distinct-oom
branch
from
August 27, 2026 10:27
ec8f139 to
1ad9237
Compare
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 17012 ms |
Contributor
TPC-DS: Total hot run time: 84162 ms |
Contributor
ClickBench: Total hot run time: 17.49 s |
Contributor
FE Regression Coverage ReportIncrement line coverage |
Contributor
Author
|
run cloud_p0 |
Contributor
FE Regression Coverage ReportIncrement line coverage |
foxtail463
force-pushed
the
fix/cbo-agg-parent-shuffle-distinct-oom
branch
from
August 28, 2026 06:03
1ad9237 to
d6c8d18
Compare
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 17027 ms |
Contributor
TPC-DS: Total hot run time: 85522 ms |
Contributor
ClickBench: Total hot run time: 17.68 s |
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.
Refine aggregate shuffle-key reuse across property derivation,
child-property regulation, final-plan pruning, and translation.
Reuse parent shuffle keys when statistics indicate a balanced
distribution, while rejecting low-cardinality or skewed subsets and
falling back to the full aggregate keys when necessary.
Keep the safety policy consistent across planning stages while
preserving existing distribution and aggregate-stage semantics.