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

Issue 748511 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Last visit > 30 days ago
Closed: Feb 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 1
Type: Bug

Blocking:
issue 748242



Sign in to add a comment

cast_unittests VideoEncoderTest.EncodesVariedFrameSizes_3 fails on 10.13

Project Member Reported by rsesek@chromium.org, Jul 25 2017

Issue description

Chrome Version: 4b523c1060e2bc62bceb42e3d1674635cd48f2e0
OS: macOS 10.13 High Sierra

What steps will reproduce the problem?
(1) Run cast_unittests on 10.13
(2) Observe failure in VideoEncoderTest.EncodesVariedFrameSizes_3

[ RUN      ] VideoEncoderTest.EncodesVariedFrameSizes/3
../../media/cast/sender/video_encoder_unittest.cc:208: Failure
Value of: operational_status_ == STATUS_CODEC_REINIT_PENDING || operational_status_ == STATUS_INITIALIZED
  Actual: false
Expected: true
../../media/cast/sender/video_encoder_unittest.cc:226: Failure
      Expected: expected_frame_id
      Which is: F16
To be equal to: encoded_frame->frame_id
      Which is: F14
[57642:775:0725/011353.574538:19976470615263:WARNING:video_encoder_unittest.cc(245)] Platform encoder returned an empty frame @ frame_id=F14


What is the expected result?
Test passes

What happens instead?
Test fails

Please use labels and text to provide additional information.

Job: https://luci-milo.appspot.com/buildbot/chromium.fyi/Chromium%20Mac%2010.13/2
Logs: https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Fchromium.fyi%2FChromium_Mac_10.13%2F2%2F%2B%2Frecipes%2Fsteps%2Fcast_unittests%2F0%2Fstdout


For graphics-related bugs, please copy/paste the contents of the about:gpu
page at the end of this report.

 
Cc: hubbe@chromium.org m...@chromium.org
Components: -Internals>Cast Internals>Cast>Streaming

Comment 2 by sko...@chromium.org, Jul 31 2017

Owner: amp@chromium.org
Status: Assigned (was: Untriaged)
Labels: -Pri-2 Pri-1
We should try to understand what's failing here ASAP.

Comment 4 by amp@chromium.org, Aug 3 2017

Just realized this was mac specific and I don't have easy access to a mac right now.  If we need to jump on this quickly someone else may need to grab it.
Cc: ellyjo...@chromium.org
Mac CQ will be upgraded to 10.13 over the coming 3-4 weeks. Please see issue 805475 for more details. Please resolve 10.13 blockers asap. 
Cc: amp@chromium.org
Owner: ccameron@chromium.org
mac triage: ccameron@, can you take a peek at this and/or work with amp@ on it? thanks :)

Comment 7 by amp@chromium.org, Jan 31 2018

Owner: amp@chromium.org
Status: Started (was: Assigned)
Apologies for dropping this. I did eventually get a mac setup, and I just recently upgraded to 10.13.

I ran this test and it passes for me (I wasn't able to get it to fail) on my desktop.


Is there way to have the bots try this out again (whatever generated this issue in the first place) to verify?

Comment 8 by rsesek@chromium.org, Jan 31 2018

The OP links to the FYI waterfall builder that is continuously running the tests. If it's passing there, then we can probably WontFix.

Comment 9 by amp@chromium.org, Jan 31 2018

Ah, thanks for the pointers.  I'm not very familiar with the waterfall builders.

It looks like there are still recent failures (most recently yesterday):
https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.fyi%2FChromium_Mac_10.13%2F871%2F%2B%2Frecipes%2Fsteps%2Fcast_unittests_on_Intel_GPU_on_Mac_on_Mac-10.13%2F0%2Fstdout

It looks like it doesn't always fail though, such as the run within the last hour:
https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.fyi%2FChromium_Mac_10.13%2F880%2F%2B%2Frecipes%2Fsteps%2Fcast_unittests_on_Intel_GPU_on_Mac_on_Mac-10.13%2F0%2Fstdout

Is there guidance for digging into bugs that only fail on the bots sometimes?
I wonder if it's a 10.13.1 vs 10.13.2 difference. Looking at the swarming details, you can see that the one that passed ran on .2 and the one that failed ran on .1.

https://chromium-swarm.appspot.com/task?id=3b66ce32b1d74e10&refresh=10&show_raw=1
https://chromium-swarm.appspot.com/task?id=3b633d7106d1a110&refresh=10&show_raw=1

Comment 11 by amp@chromium.org, Jan 31 2018

Ah, could be.  I just updated to 10.13.2 as well.

Do we need to fix it on 10.13.1, or is it fine if it works on 10.13.2?
No, probably not. But we should make sure that is the actual root cause (e.g. consistently fails on 10.13.1 and not on 10.13.2) and that it's not just flaky on 10.13 in general.

Comment 13 by amp@chromium.org, Feb 5 2018

Ack.  I'll continue watching this to see if the failures remain only on 10.13.1 for another week or so and then close it out if we don't see any on 10.13.2.
Thanks. It looks like in issue 808760 all the bots are going to get upgraded to 10.13.2 soon.

Comment 15 by amp@chromium.org, Feb 6 2018

Update after tracking this for one day.

All failures happened on 10.13.1 and all successes happened on 10.13.2.

In other words, 10.13.2 always passes and and 10.13.1 always fails.

Per issue 808760 it looks like there are no more 10.13.1 bots (the upgrade just finished and I checked a few of the bots that had failed recently and they were all now 10.13.3).

It looks to be pretty clear it was an OS issue which has been resolved with the latest updates, but I'll wait until I see a pass on 10.13.3 before closing this out.

Comment 16 by amp@chromium.org, Feb 9 2018

Status: Fixed (was: Started)
Closing this out.  Last failure was on Feb 5 just before the bots were upgraded.

Every run of this tests has passed since then.

Sign in to add a comment