This is very weird; the image is not entirely wrong, only the green color portion disappears.

I couldn't reproduce this error on another Win 10 machine. DCHECK shouldn't be a problem because the same DCHECK error happens in some other passed tests as well.

Are you sure build 20 is the first build that this flakiness start to appear? The only way I can think of identifying a culprit is by looking at the list of commits happening during that time period.

junov@: Any insights that come to your mind when you see the picture?

What I've meant by build 20 is that this goes way back, so we probably don't have the history of when it started, and should just assume it was flaky forever.

Seems like this is going to take a while to investigate, so I'm going to suppress the test.

ynovikov@: Thanks. But could you suppress it only on this particular machine? As the test was running fine in other machines.

Yeah, I'll suppress only where necessary.
Also happens in Debug:
https://build.chromium.org/p/chromium.gpu.fyi/builders/Win10%20Debug%20%28Intel%20HD%20530%29/builds/135
https://build.chromium.org/p/chromium.gpu.fyi/builders/Win10%20Debug%20%28Intel%20HD%20530%29/builds/37

Some analysis:

1. I don't think this flakiness is related to resizing, not only because the size of red-color portion is correct, but also because another resizing test on webgl context does not exhibit flakiness behavior.

2. Test was created in Dec 21 but flakiness starts to appear in Jan 13 and is more frequent on Release machine than the Debug machine. There was an update of gpu reference image in Jan 12; but that update was to make the edge between green and red portion softer, not to change its size or color. And the edge in generated image here is indeed correctly softened.

For somehow, the image rendered on the screen only shows the second "fillRect" and not the first "fillRect". Not sure if this is a color-rendering problem of green pixels on Win10 Intel HD 530 specifically (+ccameron: maybe you can have some insights on the color display here? I saw quite some color-related CLs committed by you on Jan 12-13).

For now, I'm going to try landing a patch that makes use of the returned promise of commit() to see if that can solve this weird partial rendering issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e6db2dafe177bcc8e3e22612c39b25bead34f01b

commit e6db2dafe177bcc8e3e22612c39b25bead34f01b
Author: ynovikov <ynovikov@chromium.org>
Date: Fri Feb 10 21:31:20 2017

Mark Pixel_OffscreenCanvas2DResizeOnWorker Flaky on Win10 Intel HD 530

BUG= 690663 
TBR=kbr@chromium.org, xlai@chromium.org
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2691523003
Cr-Commit-Position: refs/heads/master@{#449740}

[modify] https://crrev.com/e6db2dafe177bcc8e3e22612c39b25bead34f01b/content/test/gpu/gpu_tests/pixel_expectations.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/9ff9600d38e624c3caf86e2e8e4d1730ab7a34c8

commit 9ff9600d38e624c3caf86e2e8e4d1730ab7a34c8
Author: xlai <xlai@chromium.org>
Date: Fri Feb 10 21:55:32 2017

Improve Pixel_OffscreenCanvas2DResizeOnWorker test

This CL is an attempt to fix flaky Pixel_OffscreenCanvas2DResizeOnWorker on
Win10 Intel HD 530.

TBR=junov@chromium.org
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG= 690663 

Review-Url: https://codereview.chromium.org/2690663002
Cr-Commit-Position: refs/heads/master@{#449746}

[modify] https://crrev.com/9ff9600d38e624c3caf86e2e8e4d1730ab7a34c8/content/test/data/gpu/pixel_offscreenCanvas_2d_resize_on_worker.html

The test is still flaky: https://build.chromium.org/p/chromium.gpu.fyi/builders/Win10%20Release%20%28Intel%20HD%20530%29/builds/187

It runs two times before the image matched.

http://chromium-browser-gpu-tests.commondatastorage.googleapis.com/view_test_results.html?a537597237c4cb31028478d185e394784b961ec6_Win10_Release_Intel_HD_530__telemetry

 Issue 809884  has been merged into this issue.

Detected 3 new flakes for test/step "gpu_tests.pixel_integration_test.PixelIntegrationTest.Pixel_OffscreenCanvas2DResizeOnWorker". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNyZgsSBUZsYWtlIltncHVfdGVzdHMucGl4ZWxfaW50ZWdyYXRpb25fdGVzdC5QaXhlbEludGVncmF0aW9uVGVzdC5QaXhlbF9PZmZzY3JlZW5DYW52YXMyRFJlc2l6ZU9uV29ya2VyDA. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).

xlai@, should it be disabled on Win?

vasilii@: The test was already disabled on Win (https://cs.chromium.org/chromium/src/content/test/gpu/gpu_tests/pixel_expectations.py?l=77). I checked up the 3 flaky cases that were posted 46 hours ago, and saw that it is a mass failure on all the pixel tests with an error message "browser not started". So, the new phenomenon is not related to this test case specifically.

I manually checked the test results (because of info unavailability of flaky test crbug.com/783263) for the latest 11 builds in Win 10 Release Intel HD 630:
https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Win10%20FYI%20Release%20%28Intel%20HD%20630%29
(Cannot find HD 530 bot; maybe they've upgraded it to 630 recently).
and find that this test is successful all the time.

The previous failure might be actually a device-specific issue that is not seen in Intel HD 630. I would like to enable this test; if it goes flaky again, we can reopen this issue.

Yes, the Intel HD 530 bots have been decommissioned and replaced by a larger number of Intel HD 630 bots.

Thanks for following up on this.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7dae0564c3288a89d2c2af19dcaf169c15ffe5a7

commit 7dae0564c3288a89d2c2af19dcaf169c15ffe5a7
Author: xlai <xlai@chromium.org>
Date: Fri May 04 20:20:37 2018

Enable flaky resize OffscreenCanvas test on Win 10 Intel

The previous flakiness might be a device-specific issue that is seen in Intel
HD 530 but not Intel HD 630.

Bug:  690663 
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I18b2c9822bf39cfbcd233e2219ac460eb51875b6
Reviewed-on: https://chromium-review.googlesource.com/1044656
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Olivia Lai <xlai@chromium.org>
Cr-Commit-Position: refs/heads/master@{#556173}
[modify] https://crrev.com/7dae0564c3288a89d2c2af19dcaf169c15ffe5a7/content/test/gpu/gpu_tests/pixel_expectations.py

The Win 10 Intel HD 630 does not show flakiness of this test for 50+ builds. I'll hence mark this as fixed.

Thank you for following up on this Olivia!