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

Issue 650762 link

Starred by 5 users

Issue metadata

Status: Archived
Owner: ----
Closed: Aug 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug-Regression



Sign in to add a comment

Fix and reenable smoothnes.top_25_smooth on Win Perf bots

Project Member Reported by sullivan@chromium.org, Sep 27 2016

Issue description

I'm still poking through revision ranges to see what changed. It may be a WPR failure with expired date/time?

Gmail page failing with this error:
C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\telemetry\telemetry\internal\image_processing\_bitmap.py:152: UserWarning: Using pure python png decoder, which could be very slow. To speed up, consider installing numpy & cv2 (OpenCV).
  'Using pure python png decoder, which could be very slow. To speed up, '
Traceback (most recent call last):
  File "C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\telemetry\telemetry\internal\story_runner.py", line 86, in _RunStoryAndProcessErrorIfNeeded
    state.RunStory(results)
  File "C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\common\py_trace_event\py_trace_event\trace_event_impl\decorators.py", line 52, in traced_function
    return func(*args, **kwargs)
  File "C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\telemetry\telemetry\page\shared_page_state.py", line 301, in RunStory
    self._current_page.Run(self)
  File "C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\telemetry\telemetry\page\__init__.py", line 99, in Run
    self.RunPageInteractions(action_runner)
  File "C:\b\c\b\Win_10_Perf__3_\src\tools\perf\page_sets\top_25_smooth.py", line 69, in RunPageInteractions
    'window.__scrollableElementForTelemetry != null')
  File "C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\telemetry\telemetry\internal\actions\action_runner.py", line 220, in WaitForJavaScriptCondition
    self._tab.WaitForJavaScriptExpression(condition, timeout_in_seconds)
  File "C:\b\c\b\Win_10_Perf__3_\src\third_party\catapult\telemetry\telemetry\internal\browser\web_contents.py", line 136, in WaitForJavaScriptExpression
    e.message + '\n' + debug_message)
TimeoutException: Timed out while waiting 60s for IsJavaScriptExpressionTrue.
Console output:


INFO:root:Chrome version: 2868
[  FAILED  ] https://mail.google.com/mail/ (79140 ms)
https://build.chromium.org/p/chromium.perf/builders/Win%2010%20Perf%20%283%29/builds/4062/steps/smoothness.top_25_smooth/logs/stdio

Screenshot: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/profiler-file-id_1-2016-09-21_22-09-3576161.png

Link to failing step log:
https://build.chromium.org/p/chromium.perf/builders/Win%2010%20Perf%20%283%29/builds/4062/steps/smoothness.top_25_smooth/logs/stdio
https://build.chromium.org/p/chromium.perf/builders/Win%2010%20Perf%20%283%29?numbuilds=200
https://build.chromium.org/p/chromium.perf/builders/Win%2010%20Perf%20%283%29/builds/4062
https://build.chromium.org/p/chromium.perf/builders/Win%208%20Perf%20%283%29


If the test is disabled, please downgrade to Pri-2.

 
Note to self: check if date/time of first failure are the same, to see if it's a problem with WPR certificate expiration.

I'm still working out whether revision ranges for first failure overlap.
I started digging into this. Here are all the charts that are no longer getting data:
https://chromeperf.appspot.com/report?sid=f71a05120543ccf666b46aec319e0f63062727717a073ac6bdae838f72e97e11

They all stop at different times, revisions:
chromium-rel-win7-gpu-nvidia: 2016-09-21T23:08:48, 420107-420168
chromium-rel-win7-gpu-ati: 2016-09-21T23:11:37, 420107-420168
chromium-rel-win7-x64-dual: 2016-09-25T16:28:58, 420860-420862
chromium-rel-win10: 2016-09-26T11:03:25, 420876-420883
chromium-rel-win7-gpu-intel: 2016-09-26T16:51:46, 420888-420922
chromium-rel-win7-dual: 2016-09-27T15:14:51, 421180-421229

I'm not sure if there are different failures here. I'm going to try kicking off return code bisects on some of the different breakages.
Cc: nednguyen@chromium.org
Ned, any ideas what could be going on here? The revision ranges and timings don't overlap.
No idea. I think we need to run this locally to debug.
Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Sep 27 2016


===== BISECT JOB RESULTS =====
Status: failed


===== TESTED REVISIONS =====
Revision         Mean    Std Dev  N  Good?
chromium@420107  16.733  0.0      1  good

Bisect job ran on: winx64ati_perf_bisect
Bug ID: 650762

Test Command: src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --also-run-disabled-tests smoothness.top_25_smooth
Test Metric: mean_frame_time/https___mail.google.com_mail_
Relative Change: None
Score: 0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/winx64ati_perf_bisect/builds/1575
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9000329574983967504


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6451222150643712

| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Tests>AutoBisect.  Thank you!
Project Member

Comment 8 by 42576172...@developer.gserviceaccount.com, Sep 28 2016


===== BISECT JOB RESULTS =====
Status: completed


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N   Good?
chromium@420107  1          N/A      20  good
chromium@420169  1          N/A      20  bad

Bisect job ran on: winx64nvidia_perf_bisect
Bug ID: 650762

Test Command: src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --also-run-disabled-tests smoothness.top_25_smooth
Test Metric: mean_frame_time/https___mail.google.com_mail_
Relative Change: 0.00%
Score: 0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/winx64nvidia_perf_bisect/builds/1880
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9000329613919973056


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6383418844119040

| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Tests>AutoBisect.  Thank you!
I accidentally did a regression bisect instead of a return-code bisect in #7, but it looks like the test is flaky: for r 420107  it passed once then failed 4 times.
https://build.chromium.org/p/tryserver.chromium.perf/builders/winx64ati_perf_bisect/builds/1575


Ned, I want to debug this on a windows laptop on canary. What's the fastest way to get telemetry and run the test (and possibly re-record the page)? Is there still a zip file of telemetry ToT or do I need to sync all of Chrome source?
Cc: vmi...@chromium.org
cc vmiura, test owner. it looks like I won't be able to debug in a short time frame, so disabling this test on windows.
Victor, would you prefer to disable the benchmark on windows, or the gmail page on all platforms? (those are the knobs we have for failing tests)
Annie: to quickly grab a copy of "telemetry + perf benchmark + chromium build", you can pick any recent swarming task of telemetry_perf_unittest, then fetch the isolate.

For example, here I pick one from chromium_rel_ng build:
https://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_ng/builds/301371

Click on any shard in telemetry_perf_unittest. Here I pick shard 0:
https://chromium-swarm.appspot.com/user/task/318af5b940b27110

Then there is the instruction in the swarming page for downloading the isolate:

"Download inputs files into directory foo:
  python isolateserver.py download -I https://isolateserver.appspot.com --namespace default-gzip -s a6e26ba2135597c3da8a33937995ae51b9364f4a --target foo"



The isolateserver.py script can be found in https://github.com/luci/luci-py/blob/master/client/isolateserver.py 
If you disable on gmail, please cc the chromium perf-sheriff list since it will likely change the performance characteristics of the benchmark.
Cc: benjhayden@chromium.org aiolos@chromium.org
 Issue 649975  has been merged into this issue.
 Issue 650066  has been merged into this issue.
 Issue 650038  has been merged into this issue.
 Issue 650039  has been merged into this issue.
 Issue 649978  has been merged into this issue.
Status: Fixed (was: Untriaged)
Status: Available (was: Fixed)
Just noticed this is still a problem for many Windows bots. Reopen now.
smoothness.top_25_smooth.reference is still failing. Should we disable it on reference build until it is fixed?
Disable CL is up: https://codereview.chromium.org/2402693002/
Labels: -Pri-1 Pri-2
Summary: Fix and reenable smoothnes.top_25_smooth on Win Perf bots (was: smoothnes.top_25_smooth failing on almost all windows bots on chromium.perf)
Cc: jasontiller@chromium.org
 Issue 656704  has been merged into this issue.
Cc: majidvp@chromium.org tdres...@chromium.org sullivan@chromium.org
 Issue 652731  has been merged into this issue.
Status: Archived (was: Available)

Sign in to add a comment