The 1st roll just attempted to roll r304adc9832be7bbb3dde03c899347be3f05af952 and failed on the android bots:
https://chromium-review.googlesource.com/c/chromium/src/+/1276465

So that's likely the cause of the android failures. Then measurements.rendering_unittest.RenderingUnitTest.testRendering started failing on the other bots when r0480f8caeb27101608c99e4df2f686d6697551f4 was picked up (which seems culpable given that it touches *render* files.)

So revert those two?

Yeah, what Ben said.

I reverted the first change in https://chromium-review.googlesource.com/c/catapult/+/1277779

The failure is here (telemetry_perf_unittests on Android device Nexus 5X arm64).

[1011/202929.739110:FATAL:library_prefetcher.cc(323)] Check failed: IsOrderingSane(). 

An orderfile isn't created for arm64 IIUC. I'll look into that under a separate bug as it's separate from the measurement problem.

Created  crbug.com/894748  for android arm64 failure.

A fix is up for review: https://chromium-review.googlesource.com/c/chromium/src/+/1278850

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/395e62f6273eea18e1bd3f6689bfcd5622bcc261

commit 395e62f6273eea18e1bd3f6689bfcd5622bcc261
Author: Ehsan Chiniforooshan <chiniforooshan@chromium.org>
Date: Fri Oct 12 16:08:15 2018

Telemetry: merge running stats in testing results

Bug:  894622 
Change-Id: I89647c29faccb5654000a98cebabea92d9511618
Reviewed-on: https://chromium-review.googlesource.com/c/1278850
Reviewed-by: Ned Nguyen <nednguyen@google.com>
Commit-Queue: Ned Nguyen <nednguyen@google.com>
Commit-Queue: Ehsan Chiniforooshan <chiniforooshan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#599238}
[modify] https://crrev.com/395e62f6273eea18e1bd3f6689bfcd5622bcc261/tools/perf/measurements/rendering_util.py

Most tests pass now, except for measurements.smoothness_unittest.SmoothnessUnitTest.testSmoothness on Win7. I suspect that is the same issue as crbug.com/713260 (it's complaining about the same thing, i.e. frame_times does not exist in the results):

  Traceback (most recent call last):
    File "e:\b\s\w\ir\tools\perf\measurements\smoothness_unittest.py", line 70, in testSmoothness
      self.assertGreater(stat['frame_times'].mean, 0)
  KeyError: 'frame_times'


My CL is somehow exposing that bug on windows, too. Ned, should I disable that test on windows? It's already disabled on CrOS and linux. Or, should I revert?

Fine to disable that test since we gonna delete smoothness anyway

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0115531f0b5be3db1c565dc7a437c24fe7c17862

commit 0115531f0b5be3db1c565dc7a437c24fe7c17862
Author: Ehsan Chiniforooshan <chiniforooshan@chromium.org>
Date: Fri Oct 12 21:52:42 2018

Telemetry: disable testSmoothness on win

Bug:  894622 
Change-Id: I26e56c110a35ec21ba34ddc3064a2832184c9c96
Reviewed-on: https://chromium-review.googlesource.com/c/1277690
Reviewed-by: Ned Nguyen <nednguyen@google.com>
Commit-Queue: Ehsan Chiniforooshan <chiniforooshan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#599367}
[modify] https://crrev.com/0115531f0b5be3db1c565dc7a437c24fe7c17862/tools/perf/measurements/smoothness_unittest.py