New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 614474 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: May 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 1
Type: Bug



Sign in to add a comment

telemetry_UnitTests failing on chrome os

Project Member Reported by steve...@chromium.org, May 24 2016

Issue description

Failing builder:

https://build.chromium.org/p/chromiumos.chromium/builders/amd64-generic-telemetry/builds/8973

Also failing on x86-generic.

Snippet from client.0.DEBUG:

05/24 02:17:58.486 INFO |run_chromeos_tests:0052| [143/1027] telemetry.internal.actions.action_runner_unittest.ActionRunnerTest.testWaitForElement failed unexpectedly 27.1144s:
05/24 02:17:58.487 INFO |run_chromeos_tests:0052|   Traceback (most recent call last):
05/24 02:17:58.487 INFO |run_chromeos_tests:0052|     File "/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/testing/tab_test_case.py", line 23, in setUp
05/24 02:17:58.488 INFO |run_chromeos_tests:0052|       self._tab = self._browser.tabs[0]
05/24 02:17:58.489 INFO |run_chromeos_tests:0052|     File "/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/internal/browser/tab_list.py", line 18, in __getitem__
05/24 02:17:58.489 INFO |run_chromeos_tests:0052|       return self._tab_list_backend.__getitem__(index)
05/24 02:17:58.490 INFO |run_chromeos_tests:0052|     File "/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend_list.py", line 64, in __getitem__
05/24 02:17:58.490 INFO |run_chromeos_tests:0052|       return self.GetBackendFromContextId(context_id)
05/24 02:17:58.491 INFO |run_chromeos_tests:0052|     File "/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend_list.py", line 78, in GetBackendFromContextId
05/24 02:17:58.491 INFO |run_chromeos_tests:0052|       raise e
05/24 02:17:58.492 INFO |run_chromeos_tests:0052|   DevtoolsTargetCrashException: Devtools target crashed
05/24 02:17:58.493 INFO |run_chromeos_tests:0052|   ********************************************************************************
05/24 02:17:58.493 INFO |run_chromeos_tests:0052|   (/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend.py:341 _AddDebuggingInformation) Received a socket error in the browser connection and the tab no longer exists. The tab probably crashed.
05/24 02:17:58.494 INFO |run_chromeos_tests:0052|   ********************************************************************************
05/24 02:17:58.494 INFO |run_chromeos_tests:0052|   (/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend.py:342 _AddDebuggingInformation) Debugger url: ws://127.0.0.1:60880/devtools/page/039aa536-6488-43bd-98b1-12645f4a2b9f
05/24 02:17:58.495 INFO |run_chromeos_tests:0052|   ********************************************************************************
05/24 02:17:58.496 INFO |run_chromeos_tests:0052|   (/usr/local/telemetry/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome/tab_list_backend.py:110 _HandleDevToolsConnectionError) The browser exists and can be reached. The devtools target probably crashed.
05/24 02:17:58.496 INFO |run_chromeos_tests:0052|   Found Minidump: False
05/24 02:17:58.497 INFO |run_chromeos_tests:0052|   Stack Trace:
05/24 02:17:58.497 INFO |run_chromeos_tests:0052|   ********************************************************************************
05/24 02:17:58.498 INFO |run_chromeos_tests:0052|   	Cannot get stack trace on CrOS
05/24 02:17:58.498 INFO |run_chromeos_tests:0052|   ********************************************************************************
05/24 02:17:58.499 INFO |run_chromeos_tests:0052|   Standard output:
05/24 02:17:58.500 INFO |run_chromeos_tests:0052|   ********************************************************************************
05/24 02:17:58.500 INFO |run_chromeos_tests:0052|   	Cannot get standard output on CrOS
 
Components: Tests>Telemetry
Labels: -Pri-3 OS-Chrome Pri-1
Cc: dgozman@chromium.org
Possible suspect:

https://codereview.chromium.org/1997293002
Cc: levarum@chromium.org
I don't think that's https://codereview.chromium.org/1997293002 unless some code parses console output produced from postMessage.
Status: Started (was: Untriaged)
I am working on reproducing locally now. The error looks to be "The devtools target probably crashed" so that CL caught my attention. Hopefully we will be able to bisect it quickly.

I can't reproduce this locally, but cbuildbot is succeeding / failing as expected for first two builds, will start the next six now.

I doubt the Mojo change is at fault. It only changes behavior on windows. And if it was broken, _everything_ would be failing. Also, the change has been reverted (due to Dr memory failures).
Yup, the mojo change is not in the amd64 blame list, so it's not the culprit.
I queued up a bisect but the tryserevrs were drafted for release builds so the bisect has not started yet :(

(I did conform however that #395534 passes and #395540 failed, so a chrome change is definitely the culprit)

Comment 15 by lloz...@google.com, May 25 2016

This is affecting AFDO (profile based optimizations) since the telemetry test to captuire the profile is not working.

see autoserv.DEBUG at

https://pantheon.corp.google.com/storage/browser/chromeos-autotest-results/64535807-chromeos-test/chromeos2-row3-rack9-host11/debug/


Cc: -changwan@chromium.org steve...@chromium.org abodenha@chromium.org
Owner: changwan@chromium.org
Status: Assigned (was: Started)
ATTN: changwan@

This is the failing commit: cbd74f1402d7f3d87b032ea9ee1d4d65f6fcde8c

https://codereview.chromium.org/2001983002

I have no idea how/why this is affecting the telemetry tests.

I would like to revert the CL because I have no idea how to debug it, but am re-assigning it first in case someone has any thoughts.


Success:
https://uberchromegw.corp.google.com/i/chromiumos.tryserver/builders/telemetry/builds/0

Fail:
https://uberchromegw.corp.google.com/i/chromiumos.tryserver/builders/telemetry/builds/9

No way my cl has to do with this. it's only affecting android, and not chromeos. it may just mean flakiness of the test.
Owner: steve...@chromium.org
Status: Started (was: Assigned)
I do agree that it is surprising, but I don't know what all the telemetry / catapult system does so I wasn't sure.

I went ahead and ran the test again @ a1a73208fade64c6326c70df5d93bb08753c583d (the previous commit and the first commit in the failing run) and this time it does in fact appear to be failing:

https://uberchromegw.corp.google.com/i/chromiumos.tryserver/builders/telemetry/builds/12

So there does in fact appear to be some element of flakiness or other factor, sigh.

I will try to dig a bit deeper tomorrow.
This started passing again, so either it is very flakey, or was the side effect of another bug that got fixed (perhaps the one referenced in comment #15).

I will keep an eye on it but if the builders continue to pass I will just be thankful and mark this closed.

Cc: -amistry@chromium.org haddowk@chromium.org
Cc: amistry@chromium.org
Status: WontFix (was: Started)
Last 3 builds on both telemetry builders passed, so marking WontFix.

This is unlikely to just be flake since there were a large number of independent tests that were failing. We run well over 1000 tests with each test run so even minor flakes would cause consistent failures.
Then the difference must have been the chromeos versions. The only real difference between the outputs of try run 0 and run 12 was:

trybot-x86-generic-telemetry/R53-8368.0.0-b0
vs.
trybot-x86-generic-telemetry/R53-8371.0.0-b12

(The slaves were also different but that seems less likely to be the difference)

Sign in to add a comment