Skip to content

[fix](fe) Propagate JSON properties for multi-table routine load - #67510

Open
0AyanamiRei wants to merge 1 commit into
apache:masterfrom
0AyanamiRei:fix/multi-table-routine-load-pipeline-params
Open

[fix](fe) Propagate JSON properties for multi-table routine load#67510
0AyanamiRei wants to merge 1 commit into
apache:masterfrom
0AyanamiRei:fix/multi-table-routine-load-pipeline-params

Conversation

@0AyanamiRei

Copy link
Copy Markdown
Contributor

What problem does this PR solve?

Issue Number: None

Related PR: None

Problem Summary: Multi-table routine load cannot attach a table-specific pipeline plan before the backend discovers the destination table. The backend therefore requests a per-table plan with the format, and FE rebuilds the stream-load task from the Routine Load job. That rebuild already restored json_root and strip_outer_array, but omitted jsonpaths and num_as_string. Nested JSON records were consequently planned with an empty JSON path list and filtered as invalid rows. This change restores those two properties before generating each dynamic-table pipeline plan.

Release note

Fix multi-table routine load to honor jsonpaths and num_as_string for dynamic target tables.

Check List (For Author)

  • Test

    • Regression test
    • Unit Test
      • sh run-fe-ut.sh --run org.apache.doris.nereids.load.NereidsStreamLoadTaskTest
      • DISABLE_BUILD_UI=ON ./build.sh --fe -j48
    • Manual test
    • No need to test or manual test.
  • Behavior changed:

    • No.
    • Yes. Multi-table routine load now applies configured jsonpaths and num_as_string during per-table planning.
  • Does this need documentation?

    • No.
    • Yes.

Check List (For Reviewer who merge this PR)

  • Confirm the release note
  • Confirm test cases
  • Confirm document
  • Add branch pick label

### What problem does this PR solve?

Issue Number: None

Related PR: None

Problem Summary: Multi-table routine load defers planning until the backend discovers each target table. The follow-up FE planning task restored json_root and strip_outer_array from the routine load job but omitted jsonpaths and num_as_string, so nested JSON records were planned without their configured paths and were filtered as invalid rows. Copy these properties into the per-table stream load task so the generated scan parameters match the routine load definition.

### Release note

Fix multi-table routine load to honor jsonpaths and num_as_string for dynamic target tables.

### Check List (For Author)

- Test: Unit Test and FE build
    - sh run-fe-ut.sh --run org.apache.doris.nereids.load.NereidsStreamLoadTaskTest
    - DISABLE_BUILD_UI=ON ./build.sh --fe -j48
- Behavior changed: Yes. Multi-table routine load now applies configured jsonpaths and num_as_string during per-table planning.
- Does this need documentation: No
@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?

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