[fix](load) avoid query scanner updating load counters#63781
Draft
liaoxin01 wants to merge 1 commit into
Draft
Conversation
Contributor
|
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
Contributor
There was a problem hiding this comment.
Pull request overview
This PR fixes load-quality counter accounting in BE scanners by ensuring only load scanners update RuntimeState’s load filtered/unselected row counters, preventing query-side predicate filtering from polluting INSERT/DELETE load statistics (especially when enable_profile is enabled).
Changes:
- Gate
RuntimeState::update_num_rows_load_filtered/unselected()updates behind_is_loadinScanner::_collect_profile_before_close(). - Clarify via comments that query predicate filtering must not contribute to load-quality counters.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
liaoxin01
force-pushed
the
fix-cir-20393-master
branch
from
6a52d13 to
3b09c76
Compare
Contributor
Author
|
/review |
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Review summary: no blocking issues found.
Critical checkpoint conclusions:
- Goal/test: The PR restricts load quality counter updates to scanners that are actually part of load processing. This matches the described DELETE/subquery failure mode, and the added regression enables profile and exercises DELETE with a TVF subquery predicate.
- Scope/focus: The production change is small and focused; it preserves scan profile counter updates while excluding only load-quality counters for non-load scanners.
- Concurrency/lifecycle: No new shared mutable state, threads, locks, lifecycle ownership, or static initialization concerns were introduced. The existing atomic RuntimeState counters remain used only for statistics/accounting.
- Compatibility/configuration: No protocol, storage format, persisted metadata, or configuration changes are introduced.
- Parallel paths: The central Scanner::_collect_profile_before_close path covers OlapScanner/FileScanner subclasses that delegate to it; load scanners still update load filtered/unselected counters through the existing _is_load classification.
- Error handling/data correctness: No ignored Status or visibility-version/delete-bitmap changes. The change prevents query-side predicate filtering from corrupting load success/filtered row accounting.
- Performance/observability: The added branch is trivial and not on a hot per-row path; existing scan profile counters remain collected for observability.
- Test coverage: Regression coverage was added for the profile-enabled DELETE + TVF subquery case. I did not run the regression suite in this runner.
User focus: No additional user-provided review focus was specified.
Contributor
Author
|
run buildall |
Contributor
TPC-H: Total hot run time: 31896 ms |
Contributor
TPC-DS: Total hot run time: 173476 ms |
Contributor
BE UT Coverage ReportIncrement line coverage Increment coverage report
|
liaoxin01
force-pushed
the
fix-cir-20393-master
branch
from
3b09c76 to
1568a28
Compare
liaoxin01
force-pushed
the
fix-cir-20393-master
branch
from
1568a28 to
989d36e
Compare
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
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
skip_query_scan_load_countersquery option for UPDATE/DELETE plans that are executed through the load path.http_stream.enable_profile=true.Why
For DELETE statements with subquery scans, normal scan predicate filtered rows could be added to RuntimeState load counters. The insert/delete sink and FE insert result checks can then see invalid load totals such as
0/-2and fail withInsert has too many filtered data.The first fix restricted counters to
_is_loadscanners, but that was too broad and broketest_group_commit_http_stream:INSERT ... SELECT ... FROM http_stream WHERE ...needs the non-load http_stream scan predicate to contributeNumberUnselectedRows.Test
git diff --checkDORIS_THIRDPARTY=/data/data1/liaoxin/code/doris/thirdparty ./run-regression-test.sh --run -f regression-test/suites/load_p0/http_stream/test_group_commit_http_stream.groovy(local run reached FE, but stopped before assertions because this local cluster returned401 Access denied for user root@127.0.0.1on HTTP stream load)DORIS_THIRDPARTY=/data/data1/liaoxin/code/doris/thirdparty ./run-regression-test.sh --run -f regression-test/suites/external_table_p0/tvf/test_delete_with_tvf_profile.groovy(framework built and suite started; local run was interrupted atscpFilesbecause this machine prompts forroot@BEpassword)Issue: CIR-20393