Two different analyses got the same Swarming tasks and try-jobs results some how, while they at least had different try-jobs.

https://findit-for-me.appspot.com/waterfall/failure?url=https://luci-milo.appspot.com/buildbot/chromium.win/Win%207%20Tests%20x64%20(1)/22400
https://findit-for-me.appspot.com/waterfall/failure?url=https://luci-milo.appspot.com/buildbot/chromium.win/Win7%20Tests%20(1)/65471

https://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_variable/builds/5309
https://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_variable/builds/5304

There should be a major regression. Bump to p0.

The inconsistency is because failure grouping. 

Findit has a grouping feature to group the same failure across platforms. The plan was to trigger only one try job in the group and use the result for all analyses in group.But the triggering part was disabled. Now Findit only records groups, and each analysis will trigger their own try jobs. But in result page, we still have the check for group results(results from the first failure in group). So if analyses in group have different results, the results on page will show inconsistency.

We need to rethink how we should display the result page.

 Issue 688056  has been merged into this issue.

 Issue 650385  has been merged into this issue.

 Issue 670753  has been merged into this issue.

 Issue 707861  has been merged into this issue.