xiangfu0 opened a new pull request, #20084: URL: https://github.com/apache/pinot/pull/20084
Tenant rebalance's four dry-run integration tests start a separate Pinot/Kafka cluster even though they use the same configuration as the shared hybrid suite. Run them in `SharedHybridSuite` using an isolated table/topic scenario and the existing cleanup barriers. Direct `-Dtest=TenantRebalanceIntegrationTest` execution still starts and closes its own fixture. The four test bodies remain unchanged. This also removes unused H2/query-generator setup from those REST tests. Remove the standalone selector so Tenant still runs exactly once in lane 1-b. No production code changes. ### Validation - All 32 original scoped testcase identities, statuses, and multiplicities match between baseline and candidate: zero failures, errors, skips, or retries. - Direct Tenant execution: 4/4 passed. Full suite in reversed class order: 32/32 passed with exact identity/status parity. - Static lane selection retains the same 175 source entries with no duplicates; only Tenant changes execution within lane 1-b. - Fresh JDK 25 reactor build passed; module formatting, license formatting, checkstyle, and license checks passed. Compiler warnings were checked with deprecation/warning flags; none on changed files. ### Local timing For the existing hybrid suite plus Tenant, the Surefire phase decreased from **490.980s to 405.647s**, saving **85.333s (17.4%)** in one local comparison. JVM/cluster lifecycles decrease from two to one. Dependencies, JVM settings, and production bytecode match. These are scoped local results, not a claim about full CI wall time or repeated performance measurements. ### Remaining coverage validation This PR is a draft because full production-probe equivalence is not established. Across 7,262 identical production class files, the candidate loses **329 probes across 71 classes** and gains **524 across 58 classes**; gains do not cancel losses. Tenant/rebalance feature probes match exactly. Unseeded generated queries explain some differences, but broader lifecycle/filesystem/error-path differences remain unresolved. Local evidence was collected against baseline `b58776bca94f1686b501025cd092c3104cc2cf33`. The full four-lane CI matrix and repeated timing comparisons have not been run. -- 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]
