| Thread Previous • Date Previous • Date Next • Thread Next |
Hey,The regression hunting continued today. We had multiple random application tests failing on the 3 images over the week-end. The issue was in the image from Friday.
We found that the applications doesn't seem to randomly starts even on mako and upstart-app-launch was reverted to last good-known version. We asked to modify the testing plan to include explicitly running applications AP tests to be able to catch this in the future. https://bugs.launchpad.net/upstart-app-launch/+bug/1275017 is the bug to track it upstream. They are currently debugging and working on fixing those.
With that done, image #161 was kicked in containing that fix. However, due to infrastructure issues, we don't have yet all test results, they are getting rerun. Let's hope all AP tests pass now.
However, dogfooding on that image showed another regression (click apps, when installed, are redirecting to a wrong installation slot). After some bisecting in images and then in packages, we found the unity8 version creating it (even trying versions that were never into the image). It's reported in https://bugs.launchpad.net/unity8/+bug/1275832. We reverted it and upstream is looking at fixing this.
On another note, the dialer-app crash fix didn't fix the crash on mako when running the tests. The upstream team is trying to investigating more on why they can't reproduce it locally where we get it every time on the CI infra.
Something else to look for: another regression, but it seems to be random (we couldn't all reproduce it and need more infos): https://bugs.launchpad.net/webbrowser-app/+bug/1275855.
Current plan is to kick image #162 once the unity8 reverts is in and get test results tomorrow morning. Let's cross fingers that all tests pass, no regression in dogfooding, and the latest mentioned issue isn't new.
Cheers, Didier
| Thread Previous • Date Previous • Date Next • Thread Next |
