When CQ fails on swarming, rerun in test target level, not builder or tester level manually |
|||
Issue descriptionTake this CL patch set 7 as example: https://chromium-review.googlesource.com/c/567188/7 linux_android_rel_ng failed 4 times, and none of these is related to that CL. The author had to manually kick off CQ runs. For a particular run on linux_android_rel_ng. https://build.chromium.org/p/tryserver.chromium.android/builders/linux_android_rel_ng/builds/347849 ipc_tests ran on swarming, but failed within 12 seconds, without running any tests. https://chromium-swarm.appspot.com/task?id=3798a09e5867f610&refresh=10&show_raw=1&wide_logs=true In this case, rerunning ipc_tests automatically might be a more efficient use of both our CQ infra resources, and human time.
,
Jul 28 2017
,
Jul 28 2017
You're suggesting adding additional retries at a suite level, beyond what we do at a test level and at a build level. I don't think that's the right course of action in the generic case. I'd rather us address the systemic issues that cause those failures than paper over them with retries. We know it's been a rough week or two for linux_android_rel_ng in particular and are working to address the immediate issues affecting it over in https://bugs.chromium.org/p/chromium/issues/detail?id=739899 and https://bugs.chromium.org/p/chromium/issues/detail?id=748145. +Dirk, in case you disagree |
|||
►
Sign in to add a comment |
|||
Comment 1 by wychen@chromium.org
, Jul 26 2017