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

Issue 670657 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner:
Last visit > 30 days ago
Closed: Jan 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

10.2%-38.8% regression in webrtc.stress at 433524:434584

Project Member Reported by ivoc@chromium.org, Dec 2 2016

Issue description

See graphs below.
 

Comment 3 by ivoc@chromium.org, Dec 2 2016

Owner: phoglund@chromium.org
Status: Assigned (was: Untriaged)
Patrik, I'm having trouble figuring out what could be causing these regressions in CPU utilization. It would be great if there were corresponding FYI graphs to compare with.

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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: No values were found while testing the reference range.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean  Std Dev  N  Good?
chromium@434241  N/A   N/A      0  good
chromium@434307  N/A   N/A      0  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=cpu.utilization.browser webrtc.peerconnection
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: None

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6873
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8994384604094035152


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4970136304353280

| 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!
But there is a WebRTC roll very close, in the build before:

2016-11-21 sakal@webrtc.org Correctly pass drawn frame size when layout aspect ratio is used in EglRenderer.
2016-11-21 danilchap@webrtc.org RtpPacket::payload() return rtc::ArrayView instead of raw pointer
2016-11-21 kjellander@webrtc.org iOS: Cleanup buildbot JSON files + bump iOS version to 10.0
2016-11-19 solenberg@webrtc.org Remove unused files linux.cc/.h and linuxfdwalk.c/.h.
2016-11-18 zijiehe@chromium.org Add more logging in ScreenCapturerIntegrationTest
2016-11-18 honghaiz@webrtc.org Revert of Remove unused HttpClient class. (patchset #1 id:1 of https://codereview.webrtc.org/2511883005/ )
2016-11-18 solenberg@webrtc.org Remove unused HttpClient class.
2016-11-18 solenberg@webrtc.org Remove unused dbus.cc/.h and related things.
2016-11-18 nisse@webrtc.org Move FirewallSocketServer to test code.
2016-11-18 ehmaldonado@webrtc.org Changed the way we find the ProjectRootPath.
2016-11-18 ehmaldonado@webrtc.org Modify audio_processing_unittest to use ResourcePath instead of ProjectRootPath.
2016-11-18 nisse@webrtc.org Delete WindowPicker class and subclasses.
2016-11-18 aleloi@webrtc.org Changed the interface AudioMixer::RemoveSource to have a void return type.
2016-11-18 danilchap@webrtc.org Introduce ArrayView::subview function to return portion of the original view
2016-11-18 magjed@webrtc.org Reland of Stop using hardcoded payload types for video codecs (patchset #1 id:1 of https://codereview.webrtc.org/2513633002/ )
2016-11-18 magjed@webrtc.org Revert of Move smoothing filter to common audio. (patchset #3 id:60001 of https://codereview.webrtc.org/2484153002/ )
Cc: kjellander@chromium.org
One thing that makes it really hard here is that we can't bisect into WebRTC rolls. One possible solution is to make the telemetry tests run on the WebRTC FYI bots.
Cc: robert...@chromium.org
+robertocn

Adding webrtc support to bisect shouldn't be that difficult if this is something that would be handy.
That sounds promising, we've had  bug 656974  for a while and it seems the Bisect team was overloaded in Q4. I should however try to find time to take a look at the suggestions in that bug myself...
Project Member

Comment 10 by bugdroid1@chromium.org, Dec 6 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build.git/+/900a8675c27039ff701b98d085936ef6c82bf8e9

commit 900a8675c27039ff701b98d085936ef6c82bf8e9
Author: Roberto Carrillo <robertocn@google.com>
Date: Tue Dec 06 19:55:11 2016

[bisect] Adding webrtc to bisectable repos.

Also sorting the keys of a dictionary output to make it deterministic
and not break expectations.

TBR=simonhatch@chromium.org
BUG= 670657 

Change-Id: I9839f62dbcf73935d517a3ba01e9a5abc52de540
Reviewed-on: https://chromium-review.googlesource.com/417182
Reviewed-by: Roberto Carrillo <robertocn@chromium.org>
Commit-Queue: Roberto Carrillo <robertocn@chromium.org>

[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/bisector.py
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/depot_config.py
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/basic_bisect_other_direction.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/basic_buildbot_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/basic_linux_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/basic_mac_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/basic_win32_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/basic_win64_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/failed_build.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/failed_build_inconclusive_1.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/failed_build_inconclusive_11.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/failed_buildbucket_get.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/gathering_references_no_values.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/no_repro.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/no_values.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/retest_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/return_code.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/v8_roll_bisect.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipe_modules/auto_bisect_staging/example.expected/v8_roll_bisect_bis.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipes/bisection/android_bisect_staging.expected/local_basic_recipe_basic_device.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipes/bisection/android_bisect_staging.expected/local_basic_recipe_disconnected_device.json
[modify] https://crrev.com/900a8675c27039ff701b98d085936ef6c82bf8e9/scripts/slave/recipes/bisection/android_bisect_staging.expected/local_basic_recipe_failed_device.json

Cc: jif@chromium.org
Owner: jif@chromium.org

=== PERF REGRESSION ===


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

Hi jif@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 : Explicitely pass ownership of ReadingListWebStateObserverUserDataWrapper
Author  : jif
Commit description:
  
BUG=None

Review-Url: https://codereview.chromium.org/2513403003
Cr-Commit-Position: refs/heads/master@{#433821}
Commit  : 34145bb8726255a7da92275c2bb7bc40dda1ec14
Date    : Tue Nov 22 09:22:41 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev    N   Good?
chromium@433523  1.64647  0.325265   18  good
chromium@433739  1.62341  0.119755   18  good
chromium@433793  1.62226  0.119784   18  good
chromium@433820  1.61401  0.0764671  18  good
chromium@433821  1.57866  0.0433308  5   bad    <--
chromium@433822  1.58203  0.0264355  5   bad
chromium@433824  1.574    0.0307051  5   bad
chromium@433827  1.5838   0.0370612  5   bad
chromium@433834  1.59797  0.106551   27  bad
chromium@433847  1.59887  0.0716433  18  bad
chromium@433955  1.57946  0.0285173  8   bad
chromium@434386  1.5963   0.0935493  12  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests webrtc.peerconnection
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 3.05%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6893
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8994004065807318768


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5229518405500928

| 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!

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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: No values were found while testing the reference range.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean  Std Dev  N  Good?
chromium@433559  N/A   N/A      0  good
chromium@434584  N/A   N/A      0  bad

Bisect job ran on: staging_linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=cpu.utilization.browser webrtc.stress
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 0.00%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/staging_linux_perf_bisect/builds/48
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993994189190762448


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6444132371791872

| 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!
robertocn@: thanks for https://chromium-review.googlesource.com/417182
Looking at the bisect job comment in #13 - does that mean it wasn't able to bisect into the roll for some reason?

Comment 15 by jif@chromium.org, Dec 7 2016

Owner: ivoc@chromium.org
Bisector incorrectly assigned bug to me (CL mentioned in #12 only concerns iOS, where webRTC does not run).
Assigning to ivoc@
re: #c14

The bisect is complaining there were no actual values produced, so it can't proceed. My guess is the use of --story-filter, the default value is invalid and we're not running any tests. I'll try rekicking this with no filter. If it comes back, we should log a bug against the default filter for being invalid.
Cc: simonhatch@chromium.org
Project Member

Comment 19 by 42576172...@developer.gserviceaccount.com, Dec 12 2016


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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: Bisect failed to reproduce the regression with enough confidence.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N   Good?
chromium@433559  7.3146   2.16058  21  good
chromium@434584  7.04942  1.69233  21  bad

Bisect job ran on: staging_linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests webrtc.stress
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 3.63%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/staging_linux_perf_bisect/builds/54
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993450325294056416


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5263476899971072

| 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 21 by 42576172...@developer.gserviceaccount.com, Dec 12 2016


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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: Bisect failed to reproduce the regression with enough confidence.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean     Std Dev   N   Good?
chromium@433523  1.64241  0.326819  21  good
chromium@434386  1.61779  0.188082  21  bad

Bisect job ran on: staging_linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests webrtc.peerconnection
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 1.50%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/staging_linux_perf_bisect/builds/55
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993446045747445264


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5291502366883840

| 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 23 by 42576172...@developer.gserviceaccount.com, Dec 30 2016


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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: No values were found while testing the reference range.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision  Mean  Std Dev  N  Good?

Bisect job ran on: staging_linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=cpu.utilization.browser webrtc.peerconnection
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 0.00%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/staging_linux_perf_bisect/builds/73
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8991857145763570224


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5813468148531200

| 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!
Just FYI: I fired off this job as part of  bug 656974 ... It fails in a similar way (which is expected). I was mostly checking that it worked to fire the staging bots.

According to Simon, perf bisection support is only implemented at the staging bot so far. I assume we'll get it to the real ones soon?

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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: No values were found while testing the reference range.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision  Mean  Std Dev  N  Good?

Bisect job ran on: linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=cpu.utilization.browser webrtc.stress
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 0.00%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6963
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8991568413390201568


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6409386405134336

| 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!
re: #24

Yeah looks like my guess in #17 is right, story filter is invalid and filtering out all the tests. I kicked it off again with the story filter cleared. Yeah this is only on staging right now, would ideally like to confirm it's working before pushing it to all the bots.

Filed a bug here: https://github.com/catapult-project/catapult/issues/3107
Looked a bit closer at the graph and realized the ref moved at the same time:

https://chromeperf.appspot.com/report?sid=905cf9e0cf2bcd2ffa6b3869f0d13755faf8790a431baa418d0f8e7720b07b8e

So this regression is probably a false alarm.

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


=== Bisection aborted ===
The bisect was aborted because Bisect cannot identify a culprit: Bisect failed to reproduce the regression with enough confidence.
Please contact the the team (see below) if you believe this is in error.

===== TESTED REVISIONS =====
Revision         Mean     Std Dev   N   Good?
chromium@433523  1.62654  0.377014  21  good
chromium@434386  1.59149  0.100313  21  bad

Bisect job ran on: staging_linux_perf_bisect
Bug ID: 670657

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests webrtc.peerconnection
Test Metric: cpu_utilization/cpu_utilization_browser
Relative Change: 2.15%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/staging_linux_perf_bisect/builds/74
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8991566330522805856


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5771619195158528

| 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!
ivoc: as mentioned in #29: if both the test and the ref metric moved at the same time, we should be able to close this, right?

Comment 32 by ivoc@chromium.org, Jan 5 2017

Status: WontFix (was: Assigned)
Agreed, I'll close this. 
Project Member

Comment 34 by 42576172...@developer.gserviceaccount.com, Apr 10 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : webrtc.peerconnection
  Metric       : cpu_utilization/cpu_utilization_browser


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=cpu.utilization.browser webrtc.peerconnection

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8982667472838309504

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4970136304353280


| 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 Speed>Bisection.  Thank you!

Sign in to add a comment