wirybeaver commented on issue #19485: URL: https://github.com/apache/pinot/issues/19485#issuecomment-5985575362
### End-to-end PoC evidence — AQE-PR-5 Locally verified for [wirybeaver/pinot#73](https://github.com/wirybeaver/pinot/pull/73), commit `54cd91b914`, using a **two-server test fixture with real broker/server gRPC and materialized partition reads**—not only rule/helper unit tests. | Test target | Observed result | |---|---| | Nonempty GROUP BY: adapt consumer parallelism after producer completion. | `M -> 1` consumers; every original partition handle accounted for; result rows and schema match the nonadaptive baseline. No missing worker reports. | | Empty GROUP BY: coalesce empty partitions. | One consumer, the same empty result/schema as baseline, and complete worker-report coverage. | | Physical replanning through a second, **test-only** rule. | The replacement broker plan actually executes; the removed consumer is never dispatched; unused materialized output is cleaned up. | **Result: 5 real-RPC cases passed, 0 failures**, including two fixed-parallelism staged-dispatch regression cases. [Test source: QueryRunnerTest](https://github.com/wirybeaver/pinot/blob/54cd91b914e073c3d18bd1c41b37052594dbc5bf/pinot-query-runtime/src/test/java/org/apache/pinot/query/runtime/queries/QueryRunnerTest.java#L102-L209). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
