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

Issue 660048 link

Starred by 3 users

Issue metadata

Status: Duplicate
Merged: issue 610424
Owner: ----
Closed: Nov 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 1
Type: Bug



Sign in to add a comment

rasterize_and_record_micro.key_silk_cases failing on linux perf bots

Project Member Reported by rnep...@chromium.org, Oct 27 2016

Issue description

Project Member

Comment 2 by 42576172...@developer.gserviceaccount.com, Oct 27 2016


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


===== SUSPECTED CL(s) =====
Subject : Add serialization/deserialization for FakePictureLayer for testing.
Author  : xingliu
Commit description:
  
Previously we serialize/deserialize FakePictureLayer into PictureLayer.
This introduced an issue in LayerTreeHostTestReadyToActivateNonEmpty and
other unit test cases in layer_tree_host_unittests.cc that downcasting
from PictureLayerImpl to FakePictureLayerImpl will fail.

The data flow was:
FakePictureLayer==>PictureLayer==>PictureLayerImpl==>
FakePictureLayerImpl(downcast failure).

This CL adds procedure to process FakePictureLayer, so it can be
deserialized and down cast correctly.

The data flow becomes:
FakePictureLayer==>FakePictureLayer==>FakePictureLayerImpl

BUG= 657871 
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Review-Url: https://codereview.chromium.org/2451913002
Cr-Commit-Position: refs/heads/master@{#427886}
Commit  : b67229217f5b54fad888c74f6528be4af51e5a6e
Date    : Thu Oct 27 00:10:32 2016


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N  Good?
chromium@427874  0          N/A      2  good
chromium@427880  0          N/A      2  good
chromium@427883  0          N/A      2  good
chromium@427885  0          N/A      2  good
chromium@427886  1          N/A      2  bad    <--
chromium@427898  1          N/A      2  bad
chromium@427921  1          N/A      2  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: pixels_rasterized/pixels_rasterized
Relative Change: Zero to non-zero
Score: 0.0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6801
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997634694350438256


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5271682984443904

| 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: xingliu@chromium.org
Cc: khushals...@chromium.org
Owner: ----
Can't possibly be that patch. The code touched by it is not used in any Chrome build.
Started another bisect, hopefully it can catch the real culprit.
Project Member

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


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


===== SUSPECTED CL(s) =====
Subject : memory coordinator: Use free memory to guess available memory on Linux
Author  : bashi
Commit description:
  
Though there will be eaisly discardable memory such as buffers and
caches, don't count them as "available" memory because discarding
them will affect the overall performance of the system.

BUG= 617492 

Review-Url: https://codereview.chromium.org/2448923004
Cr-Commit-Position: refs/heads/master@{#427880}
Commit  : 8e656e05f77fb80739dc80bdfa1abfce1f45173d
Date    : Wed Oct 26 23:41:46 2016


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N  Good?
chromium@427870  0          N/A      4  good
chromium@427878  0          N/A      4  good
chromium@427879  0          N/A      4  good
chromium@427880  1          N/A      4  bad    <--
chromium@427882  1          N/A      4  bad
chromium@427885  1          N/A      4  bad
chromium@427900  1          N/A      4  bad
chromium@427930  1          N/A      4  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: pixels_rasterized/pixels_rasterized
Relative Change: Zero to non-zero
Score: 0.0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6803
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997618233885503744


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5846541405782016

| 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!
Cc: -khushals...@chromium.org
Owner: bashi@chromium.org
+bashi, do you mind taking a look.

Comment 10 by bashi@chromium.org, Oct 28 2016

Owner: ----
Thanks for triage but my CL won't affect rasterize_and_record_micro.key_silk_cases. I touched code which is only enabled when --enable-features=MemoryCoordinator is specified.
I started another bisect, if it comes back the same I'm going to do a speculative revert and see if your cl is somehow interacting in strange ways.
Cc: bashi@chromium.org
CCing Bashi so they can actually see my message above.
Project Member

Comment 14 by 42576172...@developer.gserviceaccount.com, Oct 28 2016

Cc: benjhayden@chromium.org
Owner: benjhayden@chromium.org

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

Hi benjhayden@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


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


===== SUSPECTED CL(s) =====
Subject : Relax RelatedHistogramBreakdown restrictions.
Author  : benjhayden
Commit description:
  
Currently, RelatedHistogramBreakdown keys must be suffixes of the referenced
Histogram names.
This would be annoying for memoryMetric, so this CL relaxes that constraint.

In order to prevent the usability issue of clicking on a link that says X only
to find that a histogram named Y is selected, this a title tooltip is shown
when hovering on the legend key containing the full name of the referenced
Histogram.

BUG=catapult:#2933

Review-Url: https://codereview.chromium.org/2452983003
Commit  : b1fcad56cd3362b942982ff8db6debb117d5005e
Date    : Wed Oct 26 23:09:35 2016


===== TESTED REVISIONS =====
Revision                             Exit Code  Std Dev  N  Good?
chromium@427865                      0          N/A      2  good
chromium@427900                      0          N/A      2  good
chromium@427900,catapult@bd5139fa07  0          N/A      2  good
chromium@427900,catapult@b1fcad56cd  1          N/A      2  bad    <--
chromium@427900,catapult@eeeba06a98  1          N/A      2  bad
chromium@427901                      1          N/A      2  bad
chromium@427902                      1          N/A      2  bad
chromium@427903                      1          N/A      2  bad
chromium@427905                      1          N/A      2  bad
chromium@427909                      1          N/A      2  bad
chromium@427918                      1          N/A      2  bad
chromium@427935                      1          N/A      2  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: pixels_rasterized/pixels_rasterized
Relative Change: Zero to non-zero
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6804
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997553806056924288


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5296895381995520

| 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!
Cc: -bashi@chromium.org
I'm going to kick off another bisect to confirm since this keeps on getting different results.
Cc: rnep...@chromium.org
Owner: ----
Project Member

Comment 18 by 42576172...@developer.gserviceaccount.com, Oct 28 2016


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


===== SUSPECTED CL(s) =====
Subject : Fix RenderFrameHost's Mojo objects not being recreated after a process crash.
Author  : jam
Commit description:
  
This fixes flakiness in
PolicyTest.ExtensionInstallBlacklistSharedModules
with PlzNavigate.

BUG= 504347 
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2454833002
Cr-Commit-Position: refs/heads/master@{#427848}
Commit  : fc5ee5bae3a6c40e46a28a3e2410c2170745eb5d
Date    : Wed Oct 26 22:23:41 2016


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N  Good?
chromium@427847  0          N/A      5  good
chromium@427848  1          N/A      5  bad    <--
chromium@427849  1          N/A      5  bad
chromium@427851  1          N/A      5  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: pixels_rasterized/pixels_rasterized
Relative Change: Zero to non-zero
Score: 0.0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6805
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997535302420391936


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5855824440721408

| 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!
Starting another one since every one is getting a different one.
Project Member

Comment 21 by 42576172...@developer.gserviceaccount.com, Oct 28 2016


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


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N  Good?
chromium@427840  0          N/A      5  good
chromium@427855  0          N/A      5  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter='list.animation.simple\.html' rasterize_and_record_micro.key_silk_cases
Test Metric: pixels_rasterized/list_animation_simple.html
Relative Change: 0.00%
Score: 0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6807
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997520292183478160


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5893437616816128

| 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!
Looks like this is a timeout that occurs on different test cases.  Sometimes list-recycle-transform.html, sometimes list_animation_simple.html.  First bot failure was in 427875:427920.

I'll try looking for indication that the test started taking longer to complete...
Cc: -benjhayden@chromium.org vmp...@chromium.org
Issue 660328 has been merged into this issue.
 Issue 656671  has been merged into this issue.
I believe there's been flakiness here for awhile - seeing intermittent failures of this form back to Oct 9.  There's also evidence of an increase in record time variability around Oct 4: https://chromeperf.appspot.com/report?sid=27d8972eb23851c88a12da741370ee1bea289d80de06958f78000dcf5ce14021&start_rev=412724&end_rev=427874

I've kicked off both a larger return code bisect (not sure if num_repeats will work there for flaky failures) as well as a stddev bisect for where the variability seems to go up.  Worst case and there's no obvious culprit then I'll disable the test and assign this to the test owner (vmpstr) for follow-up.
Blockedon: 610424
Mergedinto: 610424
Status: Duplicate (was: Untriaged)
Oh, it's also been disabled since May on Mac, Windows and Android due to flakiness -  issue 610424 .  Perhaps nobody cares about the this test anymore?
I'll disable on linux against that bug as well.
Blockedon: -610424

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


===== TESTED REVISIONS =====
Revision         Exit Code  Std Dev  N   Good?
chromium@422335  1          N/A      20  good
chromium@427600  1          N/A      20  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: record_time/record_time
Relative Change: 0.00%
Score: 0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6811
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997176356685694416


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5883427201482752

| 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!
Cc: bsalo...@google.com
Owner: bsalo...@google.com

=== Auto-CCing suspected CL author bsalomon@google.com ===

Hi bsalomon@google.com, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


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


===== SUSPECTED CL(s) =====
Subject : Move clip CTM application to SkRasterClip and SkClipStack
Author  : Brian Salomon
Commit description:
  
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2866

Change-Id: I914a57d6ba128acc457e12586c99ba6766eb940c
Reviewed-on: https://skia-review.googlesource.com/2866
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit  : a3b45d4f7db953472df4f11ab1595964b65175f9
Date    : Mon Oct 03 19:32:07 2016


===== TESTED REVISIONS =====
Revision                         Mean       Std Dev      N  Good?
chromium@422347                  0.0287103  0.00277566   5  good
chromium@422550                  0.0318621  0.00117004   4  good
chromium@422601                  0.0279448  0.00166406   5  good
chromium@422626                  0.0313655  0.00178456   5  good
chromium@422626,skia@14f984bc6b  0.0321724  0.00176621   5  good
chromium@422626,skia@a3b45d4f7d  0.0349241  0.000660775  5  bad    <--
chromium@422626,skia@afb48b6227  0.0347862  0.000375688  5  bad
chromium@422626,skia@f85d2a4fa1  0.0347724  0.000387529  5  bad
chromium@422627                  0.0350276  0.00049636   5  bad
chromium@422628                  0.035469   0.000411343  5  bad
chromium@422630                  0.0351897  0.000371924  4  bad
chromium@422633                  0.0354621  0.000388295  5  bad
chromium@422639                  0.0351293  0.000157706  4  bad
chromium@422651                  0.0354621  0.000710295  5  bad
chromium@422752                  0.0361586  0.000186653  5  bad

Bisect job ran on: linux_perf_bisect
Bug ID: 660048

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests rasterize_and_record_micro.key_silk_cases
Test Metric: record_time/record_time
Relative Change: 25.94%
Score: 95.0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/6812
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8997176309238926976


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5301149962338304

| 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!
Cc: andyb...@chromium.org serg...@chromium.org
Components: Infra>Flakiness
Labels: -Performance-Sheriff-BotHealth
Owner: katthomas@chromium.org
Status: Assigned (was: Duplicate)
Because of the flake, I'm re-opening this and assigning to Flakiness to see if there's a way we can improve our tools to make it easier for people to see if this was flake or not. We wasted a ton of time trying to re-run the builds, when we could've realized it was flake and disabled/fixed the test a week prior.
Owner: serg...@chromium.org
Reassigning to sergiyb, who is working on test flakiness.
Cc: katthomas@chromium.org
Owner: katthomas@chromium.org
We do not collect data from waterfall and only detect flakiness in CQ. Therefore this build was not detected or reported as flaky by chromium-try-flakes. Adding support for waterfall depends on migrating chromium-try-flakes to event_mon, which is tracked in issue 596226. Unfortunately, I am busy working on other flakiness projects atm and therefore the work in that direction is currently stalled. The new Flakiness Surface project that I work on is already based on event_mon and thus will show up waterfall flakiness for tests, but not for steps as in this case.
Status: Duplicate (was: Assigned)
Owner: ----

Sign in to add a comment