Skip to content

[fix](regression): stabilize push_down_filter_through_set_operation_with_unique_function plan shape#64516

Open
yujun777 wants to merge 1 commit into
apache:masterfrom
yujun777:fix-unique-function-set-op-shape
Open

[fix](regression): stabilize push_down_filter_through_set_operation_with_unique_function plan shape#64516
yujun777 wants to merge 1 commit into
apache:masterfrom
yujun777:fix-unique-function-set-op-shape

Conversation

@yujun777

@yujun777 yujun777 commented Jun 15, 2026

Copy link
Copy Markdown
Contributor

Problem

The regression test push_down_filter_through_set_operation_with_unique_function in nereids_rules_p0/unique_function/ is unstable across different pipeline environments.

When the cluster has only one pipeline executor (parallel=1), SplitAggWithoutDistinct skips two-phase aggregation (introduced in #63732), and hashAgg[LOCAL] is not generated. This causes the plan shape to diverge from the expected output:

Expected: ------hashAgg[LOCAL]
Actual:   ------PhysicalUnion

Fix

Add SET parallel_pipeline_task_num=2 to the test suite to lock non-single-instance mode, ensuring hashAgg[LOCAL] is always generated regardless of BE hardware.

Related PR

🤖 Generated with Claude Code

…h_unique_function shape

Set parallel_pipeline_task_num=2 to avoid single-instance optimization
(b1112e5) that skips two-phase aggregation on low-core BE clusters.
When parallel=1 on a single-BE cluster, isSingleExecutionInstance
returns true and hashAgg[LOCAL] is not generated, causing the plan
shape to diverge from the expected output.

Related PR: apache#62742

Co-Authored-By: Claude <noreply@anthropic.com>
@hello-stephen

Copy link
Copy Markdown
Contributor

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

  1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
  2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
  3. What features were added. Why was this function added?
  4. Which code was refactored and why was this part of the code refactored?
  5. Which functions were optimized and what is the difference before and after the optimization?

@yujun777 yujun777 changed the title fix(regression): stabilize push_down_filter_through_set_operation_with_unique_function plan shape [fix](regression): stabilize push_down_filter_through_set_operation_with_unique_function plan shape Jun 15, 2026
@yujun777

Copy link
Copy Markdown
Contributor Author

run buildall

@yujun777

Copy link
Copy Markdown
Contributor Author

run nonConcurrent

1 similar comment
@yujun777

Copy link
Copy Markdown
Contributor Author

run nonConcurrent

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants