Skip to content

datalake_agent: Hadoop (filesystem) catalog #2017

Description

@MisterRaindrop

Part of #2008. Letters (A, B0–B7, C, D, E) are the PRs listed there; this is B7.

Scope

  • HadoopCatalog over the volume's warehouse path (version-hint.text per table), FileIO from the volume; namespaces are directories. The hadoop server type already in the whitelist routes here.
  • Documented limit: commits rely on an atomic rename, which object stores lack, so concurrent writers to one table are unsafe on S3. Documentation only, no warning at CREATE SERVER.
  • End-to-end on local FS and on MinIO, including a test that shows what two concurrent commits do on each.

Out of scope

HDFS as storage (follow-up of A).

Depends on

B3.

Acceptance

  • Conformance suite green on local FS and MinIO; Spark opens the same warehouse path.
  • The S3 concurrent-commit behaviour is documented from the test's outcome.

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

    datalakecontrib/datalake_fdw and contrib/datalake_agent: Iceberg lake tables

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions