===== BISECT JOB RESULTS =====
Status: completed
=== Bisection aborted ===
The bisect was aborted because The metric values for the initial "good" and "bad" revisions do not represent a clear regression.
Please contact the the team (see below) if you believe this is in error.
=== Warnings ===
The following warnings were raised by the bisect job:
* Bisect failed to reproduce the regression with enough confidence.
===== TESTED REVISIONS =====
Revision Mean Std Dev N Good?
chromium@416006 27.8517 0.169082 18 good
chromium@416051 28.0116 0.245875 18 bad
Bisect job ran on: win_8_perf_bisect
Bug ID: 644370
Test Command: .\src\out\Release\performance_browser_tests.exe --test-launcher-print-test-stdio=always --enable-gpu
Test Metric: TabCapturePerformance/Capture
Relative Change: 0.10%
Score: 0
Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/win_8_perf_bisect/builds/2177
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9002246901365789728
Not what you expected? We'll investigate and get back to you!
https://chromeperf.appspot.com/bad_bisect?try_job_id=5793125996429312
| 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: completed
===== SUSPECTED CL(s) =====
Subject : Remove default begin frame sources from the renderer
Author : enne
Commit description:
This changes both cc proxies to no longer create default begin frame
sources. In cc, it must either be the case that an external begin frame
source is provided (with use_external_begin_frame_source set) or the
output surface provides a begin frame source (with
use_output_surface_begin_frame_source).
This changes the world from three states of BFS creation (default,
output surface provided, and external) to only two (output surface
provided and external). In the future, external will also be removed.
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
Committed: https://crrev.com/da025c649b0c2339771f2b06c482dcf8e388e889
Review-Url: https://codereview.chromium.org/2083423006
Cr-Original-Commit-Position: refs/heads/master@{#415757}
Cr-Commit-Position: refs/heads/master@{#416033}
Commit : 5593b84f823ee1517347cf773c6cd1854c0268ba
Date : Thu Sep 01 20:24:11 2016
===== TESTED REVISIONS =====
Revision Exit Code Std Dev N Good?
chromium@416005 0 N/A 2 good
chromium@416029 0 N/A 2 good
chromium@416032 0 N/A 2 good
chromium@416033 1 N/A 2 bad <--
chromium@416034 1 N/A 2 bad
chromium@416035 1 N/A 2 bad
chromium@416041 1 N/A 2 bad
chromium@416052 1 N/A 2 bad
chromium@416098 1 N/A 2 bad
Bisect job ran on: win_8_perf_bisect
Bug ID: 644370
Test Command: .\src\out\Release\performance_browser_tests.exe --test-launcher-print-test-stdio=always --enable-gpu
Test Metric: TabCapturePerformance/Capture
Relative Change: Zero to non-zero
Score: 99.9
Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/win_8_perf_bisect/builds/2182
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9002163112630548016
Not what you expected? We'll investigate and get back to you!
https://chromeperf.appspot.com/bad_bisect?try_job_id=5869628222341120
| 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 think I have the culprit:
Author: dtu <dtu@chromium.org>
Date: Fri Sep 2 11:52:03 2016 -0700
Clean up some discrepancies in the perf recipe configs
Remove "disable_tests" from all configs. Instead rely on "test_generators" and "test_spec_file" to determine whether generated tests should run.
Remove optional field "test_spec_file." The default is the name of the master. This means that tryserver.chromium.perf and chromium.perf.fyi will use their respective files instead of chromium.perf. Th
is is a noop as those files either are empty or don't exist.
Change bisect testers to use the "tester" config instead of "builder." This is mostly a no-op since those configs were unified in previous CLs.
Remove the "perf-id" and "results-url" properties. I don't know if they were used for anything.
BUG= 640287
Review-Url: https://codereview.chromium.org/2270563005
Looks like I just didn't let it wait long enough, and some of the bots are starting to pass the variables again.
I think that the failure found by the bisect might have been real though, and just happened to coincide with this problem unfortunately.
Comment 1 by 42576172...@developer.gserviceaccount.com
, Sep 6 2016