New issue
Advanced search Search tips

Issue 860092 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Jul 11
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 2
Type: Bug
Flaky-Test: WizardControllerDeviceStateWithInitialEnrollmentTest.ControlFlowWaitSystemClockSyncTimeout



Sign in to add a comment

WizardControllerDeviceStateWithInitialEnrollmentTest.ControlFlowWaitSystemClockSyncTimeout is Flaky

Project Member Reported by Findit, Jul 4

Issue description

Labels: -Pri-1 -Sheriff-Chromium OS-Chrome Pri-2
Owner: jdufault@chromium.org
Status: Assigned (was: Available)
Let me assign the owner to an owner of chrome/browser/chromeos/login.

The test is now been disabled, so let me deprioritized this and remove sheriff flag.
Project Member

Comment 3 by bugdroid1@chromium.org, Jul 4

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7533c1f9d913fc2929be12651a9a7333ff9dbc78

commit 7533c1f9d913fc2929be12651a9a7333ff9dbc78
Author: Makoto Shimazu <shimazu@chromium.org>
Date: Wed Jul 04 04:26:52 2018

Disable a test in WizardControllerDeviceStateWithInitialEnrollmentTest

WizardControllerDeviceStateWithInitialEnrollmentTest.ControlFlowWaitSystemClockSyncTimeout
is now flaky. It seems affected by crrev.com/c/1118532, but that CL just added
some CHECKs to check if it's not captured by ThreadTaskRunnerHandle. I think we
need to keep the task runner properly.

TBR=jdufault@chromium.org
NOTRY=true

Bug:  860092 
Change-Id: Id7e566d4a48f59e2f2b7121c0875afcd9dc09b24
Reviewed-on: https://chromium-review.googlesource.com/1125554
Reviewed-by: Makoto Shimazu <shimazu@chromium.org>
Commit-Queue: Makoto Shimazu <shimazu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#572471}
[modify] https://crrev.com/7533c1f9d913fc2929be12651a9a7333ff9dbc78/chrome/browser/chromeos/login/wizard_controller_browsertest.cc

Cc: morlovich@chromium.org
Sorry for the trouble. It might just need RunUntilIdle() or something; it looks like the Reply portion of some sort of a PostTaskAndReply trying to deliver it to the MockTestTaskRunner in the body is happening after the test body exited (building right now, so very rough analysis)

The task in question is apparently MobileConfig::LoadConfig()
... Though RunUntilIdle isn't enough since both the task scheduler pool and the current thread need to finish running, of course.


Also this one:
https://cs.chromium.org/chromium/src/chrome/browser/metrics/metrics_reporting_state.cc?rcl=e36be47c16e8074865dbbd4aaa631b749d2eefa5&l=100
... just more rare, and for some reason not helped by flushing task scheduler first.

I may just need to give up on thoss  CHECKs.

Re. "RunUntilIdle isn't enough since both the task scheduler pool and the current thread need to finish running, of course." :

content::RunAllTasksUntilIdle() will do just that FWIW.

(but I'm fine getting rid of those CHECKs)
Project Member

Comment 8 by bugdroid1@chromium.org, Jul 10

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d479013f0ea07513550c7ec7f8c81e08ad2a19d4

commit d479013f0ea07513550c7ec7f8c81e08ad2a19d4
Author: Maks Orlovich <morlovich@chromium.org>
Date: Tue Jul 10 00:49:18 2018

TestMockTimeTaskRunner: Remove just added CHECKs on posts to dead one

Turns out these are just too hard to avoid when TestMockTimeTaskRunner
is used in a browser test.

Bug:  860092 
Change-Id: I11e3dcfacf073549c56a40eb6840de1df4a52896
Reviewed-on: https://chromium-review.googlesource.com/1127930
Commit-Queue: Maks Orlovich <morlovich@chromium.org>
Reviewed-by: Jacob Dufault <jdufault@chromium.org>
Reviewed-by: Gabriel Charette <gab@chromium.org>
Cr-Commit-Position: refs/heads/master@{#573565}
[modify] https://crrev.com/d479013f0ea07513550c7ec7f8c81e08ad2a19d4/base/test/test_mock_time_task_runner.cc
[modify] https://crrev.com/d479013f0ea07513550c7ec7f8c81e08ad2a19d4/chrome/browser/chromeos/login/wizard_controller_browsertest.cc

Status: Fixed (was: Assigned)

Sign in to add a comment