The code modified in the Skia roll isn't used by Chrome.

The services/ui code (https://chromium.googlesource.com/chromium/src/+/08f982ad2a3cf6d5b0ce5fc7326e01b3f6073b10) doesn't run on android.

Going to investigate which change was it.
Note that it always fails on 5X, but flakes on other Nexuses.

Confirmed that the breaking CL is https://codereview.chromium.org/2340463003 - catapult roll.
Specifically https://codereview.chromium.org/2342503002 - Revert of Revert "[telemetry] Wire TsProxy through telemetry's network stack."

Now the failure is fixed by https://codereview.chromium.org/2344543002.
Specifically https://codereview.chromium.org/2335403003 - Reland of "[telemetry] Wire TsProxy through telemetry's network stack."

Going to add presubmit check for catapult rolls.

Yuly pointed out that he posted https://skia-review.googlesource.com/2420 for this. I don't seem to be able to CC: Catapult team members, so folks, please look at that CL directly.

Sorry we just reland the ts_proxy change again. I will take a look at why this is failing.

Please don't revert the change since it time is revert & reland, it might shift performance metrics of all perf benchmark.

I found the problem. It's due to the weirdness of of the maps's WPR archive. It uses WPR but the URLs are localhost URLs. With the ts_proxy change, anytime we see localhost URL, we map it to the local server and not WPR server.

I think the fix here is either updating the maps WPR to not use localhost URL or convert it to static html pages like webgl_conformance_test.

Actually I am not sure whether my CL is the cause of the failure. I revert 3cc40b37372927435ed69f91aec2117488b0825c on my local & the test is still fail. Also the it worths noting that this is only failing on Android & desktop machines, whereas if #8 is the cause, it should also fail on desktop.

Reassign this to ynovikov@ for investigating.

Meanwhile, https://codereview.chromium.org/2347483004/ is the speculative fix based on #8's analysis.

Ned, could you please continue to own this bug? Given when the test started failing and passing, it is almost 100% certain that it was those Telemetry changes which broke and then fixed the test again. Yuly is not really responsible for this bot; he's just been trying to add new tests to it recently.

Too late, Ken, already started looking.
ToT (chromium at 444b3d20c7af1b51db928b5ed65e9a099970d945, catapult at 3cc40b37372927435ed69f91aec2117488b0825c Revert of "[telemetry] Wire TsProxy through telemetry's network stack.") fails.
Checking out catapult at the previous commit 5148bd4b2be6fddb9fb33cae504afe76e113cfbd Fix typo with setting chrome category in TimelineBasedMeasurement passes.

Going to try https://codereview.chromium.org/2347483004/ next.

https://codereview.chromium.org/2347483004/ with TsProxy either enabled or disabled makes the test work for me locally.

An idea why the test fails for you locally, Ned - make sure the device has rotation disabled, i. e. is in portrait mode.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c582e5311da945d4e123f499be07f3bba2976750

commit c582e5311da945d4e123f499be07f3bba2976750
Author: nednguyen <nednguyen@google.com>
Date: Fri Sep 16 00:34:40 2016

Update the host URLs in WPR archives used by map test

This is done using:
./httparchive.py remap-host maps_004.wpr localhost:10020 map-test
(https://codereview.chromium.org/2346853003)

BUG= 646924 
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/2347483004
Cr-Commit-Position: refs/heads/master@{#419056}

[modify] https://crrev.com/c582e5311da945d4e123f499be07f3bba2976750/content/test/gpu/gpu_tests/maps.py
[modify] https://crrev.com/c582e5311da945d4e123f499be07f3bba2976750/content/test/gpu/page_sets/data/maps.json
[add] https://crrev.com/c582e5311da945d4e123f499be07f3bba2976750/content/test/gpu/page_sets/data/maps_004.wpr.updated.sha1

Verified that the test passes locally with the CL in #16, and also on the bots.
Also made sure that master.tryserver.chromium.android:android_optional_gpu_tests_rel would be tried in future rolls in https://skia-review.googlesource.com/2420 and https://codereview.chromium.org/2349843003/

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f4c80bde8a17c5cf78dd1bfeed9b4b2f2b7ad315

commit f4c80bde8a17c5cf78dd1bfeed9b4b2f2b7ad315
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Tue Sep 20 01:14:10 2016

Roll src/third_party/catapult/ c36cf56a4..ac4a5f451 (5 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/c36cf56a4fcc..ac4a5f45193a

$ git log c36cf56a4..ac4a5f451 --date=short --no-merges --format='%ad %ae %s'
2016-09-19 ynovikov Add a note on android_optional_gpu_tests_rel trybot to manual rolls doc.
2016-09-19 lpy Make radio picker fire an event when selected key changed.
2016-09-19 sullivan Do not cache LastAddedRevision entities in memcache by default
2016-09-19 sullivan Do not cache Row entities in memcache by default.
2016-09-19 sullivan Do not cache TestMetadata entities in memcache by default.

BUG= 646924 

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel
TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2350193004
Cr-Commit-Position: refs/heads/master@{#419621}

[modify] https://crrev.com/f4c80bde8a17c5cf78dd1bfeed9b4b2f2b7ad315/DEPS