network_service_browser_tests major source of flakes |
|||
Issue descriptionhttps://datastudio.google.com/c/reporting/12dYEpcepJ5_6ZOhprbd5GpDNooiUJONV/page/0C9Y network_service_browser_tests on Win-10 is responsible for more than 50% of all CQ flakes across all builds. This spiked sometime 9/26 or 9/27. Please turn off the test suite and investigate.
,
Sep 28
(See https://crbug.com/890380 )
,
Sep 28
Thanks for the fast follow up. Since we expect this to clear up, let's wait a day and see if numbers drop.
,
Sep 28
Turning off the test suite is a big ask, so please add more data. i.e. I don't know how this graph or data was generated. The main waterfall doesn't show any flakes in the last 30 builds. Before 30, both NS and non NS integration tests were flaky for ~100 builds. https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20Tests%20x64?limit=200 The CQ bot runs don't show NS being more flaky either. Ignoring runs with both non-NS and NS flake from, https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win10_chromium_x64_rel_ng?limit=200 I only see 2 instances where network service was flaky https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win10_chromium_x64_rel_ng/104921 https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win10_chromium_x64_rel_ng/104898 both times were because of left over files in the swarming bots, and these files aren't ones that network process uses. e.g. Failed to delete C:\b\s\w\itroincm. The following files remain: - \\?\C:\b\s\w\itroincm - \\?\C:\b\s\w\itroincm\cb7a69ae-dc5d-4a5e-a1aa-b35a65d200dd.tmp Failed to delete C:\b\s\w\itroincm. The following files remain: - \\?\C:\b\s\w\itroincm - \\?\C:\b\s\w\itroincm\cb7a69ae-dc5d-4a5e-a1aa-b35a65d200dd.tmp 7988 2018-09-28 18:40:33.030 E: rmtree(u'C:\\b\\s\\w\\itroincm') failed: [Error 32] The process cannot access the file because it is being used by another process: u'\\\\?\\C:\\b\\s\\w\\itroincm\\cb7a69ae-dc5d-4a5e-a1aa-b35a65d200dd.tmp' from the first one.
,
Dec 4
No longer major source of flakes. Closing the bug. |
|||
►
Sign in to add a comment |
|||
Comment 1 by mmenke@chromium.org
, Sep 28