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

Issue 675034 link

Starred by 1 user

Issue metadata

Status: Duplicate
Owner:
Closed: Dec 2016
Cc:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 1
Type: Bug-Regression



Sign in to add a comment

smoothness.sync_scroll.key_mobile_sites_smooth (and thread_times.key_hit_test_cases?) failure on chromium.perf all Android bots

Project Member Reported by charliea@chromium.org, Dec 16 2016

Issue description

Revision range first seen: I'm not sure. All of the bots seem to be failing after https://codereview.chromium.org/2584783002 is in the build, but I'm completely baffled as to why that CL would be problematic. Given that, on almost all of the bots, it's the only CL in the particular roll where this benchmark started failing, though, I have to imagine that it's the problem. See the attached pics.

Link to failing step log: https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_One_Perf__3_%2F4534%2F%2B%2Frecipes%2Fsteps%2Fsmoothness.sync_scroll.key_mobile_sites_smooth%2F0%2Fstdout

All of the new failures appear to be happening on http://ftw.usatoday.com/2014/05/spelling-bee-rules-shenanigans.

I'm going to revert the CL and see if this fixes the situation.

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

 
evidence_1.png
112 KB View Download
evidence_2.png
151 KB View Download
Launched a perf try job to test out the revert here: https://codereview.chromium.org/2580053002
Cc: charliea@google.com
Cc: -charliea@google.com
Summary: smoothness.sync_scroll.key_mobile_sites_smooth (and thread_times.key_hit_test_cases?) failure on chromium.perf all Android bots (was: smoothness.sync_scroll.key_mobile_sites_smooth failure on chromium.perf all Android bots)
I think that this may also be affecting thread_times.key_hit_test_cases, FWIW.
It doesn't look like the perf try job was successful in fixing this, meaning that the CL I thought might be responsible probably isn't. AFAICT, that means that there's some infrastructure change that's responsible for this, but it's not quite clear to me how to find that.
sullivan@ suggested that the bot might be providing us with an incorrect revision range. Because of that, I'm going to launch a bisect with a broadened revision range and see what it comes up with.
Project Member

Comment 8 by 42576172...@developer.gserviceaccount.com, Dec 19 2016

Cc: alexclarke@chromium.org
Owner: alexclarke@chromium.org

=== TEST FAILURE ===


=== Auto-CCing suspected CL author alexclarke@chromium.org ===

Hi alexclarke@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


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


===== SUSPECTED CL(s) =====
Subject : Dont post delayed DoWork for disabled queues. We could do this for fences too but don't
Author  : alexclarke
Commit description:
  because that would break the throttling logic which needs to know when the next delayed
task is due to be run.

BUG= 671669 

Review-Url: https://codereview.chromium.org/2572893002
Cr-Commit-Position: refs/heads/master@{#438962}
Commit  : cab9842ac55e5b9dab766f11c6e412949d854483
Date    : Fri Dec 16 00:10:17 2016


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N  Good?
chromium@438940  0          N/A      1  good
chromium@438955  0          N/A      1  good
chromium@438959  0          N/A      1  good
chromium@438961  0          N/A      1  good
chromium@438962  1          N/A      1  bad    <--
chromium@438963  1          N/A      1  bad
chromium@438970  1          N/A      1  bad
chromium@439000  1          N/A      1  bad

Bisect job ran on: android_nexus5_perf_bisect
Bug ID: 675034

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=ftw.usatoday smoothness.sync_scroll.key_mobile_sites_smooth
Test Metric: avg_surface_fps/http___ftw.usatoday.com_2014_05_spelling-bee-rules-shenanigans
Relative Change: 0.00%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus5_perf_bisect/builds/4473
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8992819288306958080


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5828482381119488

| 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!
Owner: skyos...@chromium.org
+Sami
Mergedinto: 674895
Status: Duplicate (was: Untriaged)

Sign in to add a comment