maheshrijal opened a new pull request, #20728: URL: https://github.com/apache/druid/pull/20728
Fixes #20644. ### Description `HistoricalCloningTest` started `coordinator1` and `coordinator2` together and assumed `coordinator1` would win the leader election. Start order does not guarantee that. Only the leader emits `segment/loadQueue/success`, so when `coordinator2` won, the wait on `coordinator1`'s emitter timed out after 10 s. In the failing run that wait timed out 10.26 s into the test, while the passing run on the next master commit finished the whole class in 12 s. This change adds `coordinator2` to the cluster only after `coordinator1` has loaded the 10 segments, so `coordinator1` is the only possible leader for those waits. Stopping `coordinator1` still forces a real leader switch, and the clone and turbo-load checks then run on `coordinator2`. The only difference is that `coordinator2` now starts just before it takes over instead of running from the start; its server view syncs in well under a second, before `historical2` joins. #### Release note None. Test-only change. <hr> This PR has: - [x] been self-reviewed. ## Council review Rounds: 1 | Delegate | Round 1 | | --- | --- | | Claude Opus | 9/10 | | GPT Sol | 9/10 | | Gemini (Gemini 3.8 Flash High) | 10/10 | #### Generative AI usage Written with help from Claude. -- 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]
