Skip to content

[Bug] Primary-key vector search omits fresh level-0 files in full/detail search modes #9894

Description

@LuciferYang

Search before asking

  • I searched in the issues and found nothing similar.

Paimon version

master

Compute Engine

Any (primary-key table vector search)

Minimal reproduce step

On a primary-key table with a pk-vector index and freshly-written rows that have not been compacted yet (level-0 APPEND files present in the bucket), run a vector search with vector-index.search-mode set to full or detail.

What doesn't meet your expectations?

Rows written since the last compaction are missing from the result.

full and detail are documented to exact-scan unindexed data files, but PrimaryKeyVectorRead handed the bucket search only the files PrimaryKeyIndexSourcePolicy.shouldRead accepts (complete COMPACT level>0 outputs). Level-0 APPEND files were therefore dropped by every search mode. FAST intentionally ignores unindexed files, but full/detail should scan them.

Anything else?

No response

Are you willing to submit a PR?

  • I'm willing to submit a PR!

Activity

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions