SEZ9 commented on issue #12483: URL: https://github.com/apache/seatunnel/issues/12483#issuecomment-5842784401
Thanks for the detailed write-up and code references. The observation that `RUN_ALL_CONTAINER=false, RUN_ZETA_CONTAINER=false` falls through to all 7 engines, and that the `updated-modules-integration-test-part-N` jobs never set either flag, matches the current behaviour; the routing asymmetry between connector-only and engine-only PRs is a fair point. The proposal is clear as written: an explicit `E2E_ENGINES` input with the current behaviour as default, a reduced set for connector-only PR jobs, all 7 engines kept for `api=true` and `seatunnel-translation/**` changes, the nightly as the backstop for the dropped engines, and a maintainer-applied `full-e2e` label to restore the full set on a PR. Two things would help move the dev@ discussion forward: 1. The 90-day count of connector E2E failures specific to Flink 1.13/1.15 or Spark 2.4 you mentioned. That number is the main input for deciding whether the nightly is an acceptable backstop. 2. A measurement on one connector module comparing the 7-engine run with the proposed 3- or 4-engine set, so the 40–55% estimate can be checked before any default changes. Agreed that no implementation should start before there is agreement on dev@. <!-- streview-comment:1326 --> -- 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]
