We have Jenkins voting through two ways: 1) via a review.gluster.org_for-smoke-jobs Gerrit server and Jenkins posts the results into the Smoke flag. 2) via Centos-regression and NetBSD-regression where the build nodes SSH in to post the job status. This is causing two different sets of problems 1) If you modify the patch after a regression run started, the results will be posted to the patch against with regression was started rather than the latest one. This is okay most of the time since new patch = new regression run anyway, but there are cases where Jenkins is smart: a) If you push two jobs to Jenkins (for the same patch) and neither has started, Jenkins will only start one job. b) If you push change the commit message, we theoretically retain votes. Jenkins votes on the earlier patch making it pointless. 2) If you retry smoke and regression at the same, Jenkins will start them all together and combine the reporting. If regression fails, this makes the reporting go haywire.
Lot of time since no activity on this bug. We have either fixed it already or it is mostly not critical anymore! Please re-open the bug if the issue is burning for you, or you want to take the bug to closure with fixes.