The GitHub Actions job "Build" on texera.git/main has succeeded. Run started by GitHub user Yicong-Huang (triggered by Yicong-Huang).
Head commit for run: 02018cdfc25eccde569e34a69d45bce18e8a692d / Yicong Huang <[email protected]> fix(amber): Python internal marker replay during reconfiguration (#4547) ### What changes were proposed in this PR? This PR fixes the Python reconfiguration hang reported in #4545 and explicitly re-enables the regression test that was temporarily ignored in #4546. The regression came from `#4424` (`ef66190f22`), which changed the lifetime of Python `current_internal_marker`. After that change, `get_internal_marker()` stopped consuming the marker on read, and the main loop deferred cleanup until after replaying internal channel markers. For Python source operators, that allowed an internal `EndChannel` marker to remain visible across the pause and reconfiguration window. When the reconfiguration ECM was processed, the stale marker could be observed and replayed again, which corrupted end-of-stream handling and caused the workflow to hang. This PR restores the expected one-time consumption behavior by: - making `get_internal_marker()` consume `current_internal_marker` when it is read - removing the extra delayed cleanup after replaying internal channel markers in the Python main loop - changing `should propagate reconfiguration through a source operator in workflow` in `ReconfigurationSpec` from `ignore` back to a normal enabled test ### Any related issues, documentation, discussions? Fixes #4545. Regression introduced by #4424. Re-enables the temporary test disable from #4546 after fixing the underlying lifecycle bug. ### How was this PR tested? Tested with existing Scala tests using Java 11: - `WorkflowExecutionService/testOnly org.apache.texera.amber.engine.e2e.ReconfigurationSpec` This run included the re-enabled `should propagate reconfiguration through a source operator in workflow` case. The full `ReconfigurationSpec` passed on the rebased branch (`5/5` passed, `0` ignored). ### Was this PR authored or co-authored using generative AI tooling? Generated-by: OpenAI Codex (GPT-5) Report URL: https://github.com/apache/texera/actions/runs/25048589690 With regards, GitHub Actions via GitBox
