databird07 opened a new pull request, #20363: URL: https://github.com/apache/hudi/pull/20363
### Describe the issue this Pull Request addresses Every test job in `bot.yml` runs its own `mvn clean install -DskipTests` before testing. In run 38362333401 the 11 builds that share a profile set took 63.5 of ~622 job-minutes. ### Summary and Changelog - Three build jobs (JDK11/Spark 3.5, Java 17/Spark 4.2, Flink) save their module `target/` directories and `~/.m2/.../org/apache/hudi` as a workflow artifact (pax tar + zstd, duplicate jars hard-linked; ~1 GB). - Test jobs that use the same profiles restore that artifact instead of building. Restored timestamps are re-stamped (`compress_mtimes.py`) so Maven treats the outputs as up to date. Test steps themselves are unchanged. - Bundle validation, Spark 4 bundle validation and integration tests keep their own builds because they use different profiles. ### Impact CI-only. Projected saving is ~30 job-minutes per push. The critical path gets ~3 minutes longer, because test jobs wait for a build job. Locally, 20 runs each of in-job build vs. restored artifact gave identical test counts and identical JaCoCo coverage, with 0 recompiled sources after restore. This hasn't run on GitHub Actions yet, so this PR's own CI run is the first real check. ### Risk Level medium. Workflow restructuring that depends on the restored outputs being treated as up to date. Mitigated by manifest checks (commit and profile set) on restore and by the local comparison above. Reverting is a single-commit revert. ### Documentation Update none ### Contributor's checklist - [x] Read through [contributor's guide](https://hudi.apache.org/contribute/how-to-contribute) - [x] Enough context is provided in the sections above - [x] Adequate tests were added if applicable -- 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]
