Linux ChromeOS Buildspec Tests alternates channels |
||||||||
Issue descriptionLinux ChromeOS Buildspec Tests is strange. It runs browser_tests and interactive_ui_tests on linux-chromeos, but not unit_tests, which seems odd. Furthermore, it appears to alternate between Release, Beta, and Dev (and maybe Canary?) channels, which is *super* confusing. Even if the latest build is green, it might be failing in a different channel. If this builder is useful (and it probably is, I just found what may be a legitimate problem in M50, after spending quite a while wondering why it has been alternating red and green), we should split it up by channel.
,
Apr 25 2016
Ok. I don't think chromeos infra owns this, given that it isn't even running cbuildbot.
,
Apr 25 2016
So, who who owns this then? It is part of the chromeos.chrome waterfall.
,
Apr 25 2016
This would be owned by Chrome Gardeners or Chrome infra.
,
Apr 25 2016
FYI I saw some design docs going around re. the alternating channel behavior. I believe it's intentional for Chrome-side release builders to alternate channels. Chrome OS doesn't have alternating-channel builders though.
,
Apr 25 2016
For the most part Chrome (OS) Gardeners (myself included) aren't very familiar with how builders are set up. I'm not sure why Chrome infra would own this? It is entirely Chrome OS specific and on a waterfall with other Chrome OS builders. The only reference I can find in the code to "Linux Chrome Buildspec Tests" is in browsertest_status.py which doesn't appear to be a test configuration file: https://cs.corp.google.com/chromeos_public/src/platform/crostestutils/provingground/browsertest_status.py?q=%22Linux+ChromeOS+Buildspec+Tests%22&l=32 I am just trying to document what this does and how we should garden it, but it is all very confusing.
,
Nov 11 2016
Builder is working as designed, per 'Running Chrome ‘browser_tests’ for CrOS Releases' design doc: https://docs.google.com/document/d/1IMB4K_CcZnY5GaXs-TXxWX6h6bD-01iUlXa8fNM_40Y/edit#heading=h.s4abdvxcitqs See related task bug: 447279
,
Nov 11 2016
While I acknowledge this may be working as designed, as the description says, this is extremely confusing and makes tracking down failures very difficult. For an example, see issue 654561#c32: https://bugs.chromium.org/p/chromium/issues/detail?id=654561#c32 I am re-opening and changing this over to a feature request. Given that it is generally the gardeners that have to investigate these failures, I do not think their needs should be discounted.
,
Nov 19 2016
,
Dec 22 2016
|
||||||||
►
Sign in to add a comment |
||||||||
Comment 1 by akes...@chromium.org
, Apr 25 2016