kubeflow/sdk

KEP-107: Improve Spark E2E Robustness

Aperta

#611 aperta il 19 lug 2026

 (7 commenti) (0 reazioni) (1 assegnatario)Python (196 fork)auto 404
area/sparkarea/testinggood first issuehelp wantedkind/feature

Metriche repository

Star
 (124 stelle)
Metriche merge PR
 (Metriche PR in attesa)

Descrizione

What you would like to be added?

The goal of this issue is to make the Spark E2E tests more robust and align closely with upstream artifacts to enable better detection of drifts.

The following is a consolidated list of target end states:

  • Remove the vendored SparkConnect CRD and pull all the latest upstream artifacts to test Spark SDK against the latest version.
  • Remove the default service account from the RBAC.
  • Enable "cluster watcher" support for both SparkConnect and SparkApplication.

Why is this needed?

Test robustness ensures the Spark package works in realistic production settings.

Love this feature?

Give it a 👍 We prioritize the features with most 👍

Guida contributor