1441537 - https://shaka-player-demo.appspot.com/demo/#asset=//storage.googleapis.com/shaka-demo-assets/sintel-widevine/dash.mpd stops playback 25 seconds in

Status: NEW

(Core :: Audio/Video: Playback, defect, P3)

Description

[alex_mayorga]

¡Hola!

Steps:

 - Open https://shaka-player-demo.appspot.com/demo/#asset=//storage.googleapis.com/shaka-demo-assets/sintel-widevine/dash.mpd
 - Click the white on red "Load" button

Actual result:
Playback stops at around 25 seconds.

Expected result:
Playback works normally.

¡Gracias!

Comment 1

[Ralph Giles (:rillian)]

Thanks for the report. Might be an EME issue, might be an issue with the player's remuxing. Bryce, can you please figure out what's up?

Comment 2

[Bryce Seager van Dyk [:bryce] (he/him) - Not reading bugmail]

Brain dump of investigation:

- Looks like dynamic resolution is involved.
- Once you start changing settings the original incarnation of this bug gets hard to repro. I've been clearing my shaka cookie and refreshing the player page, otherwise I find some settings are sticky.

Failure summary:

- Shaka player loads with default settings, enable adaptation will be set, no resolution for video yet.
- Upon clicking load the video will load at 426*182 and buffer some amount.
- Playback begins and adaptive streaming changes resolution setting to 1920*818 in the first few seconds, at ~13 seconds we hit our high res buffer.
- Playback continues to 24 seconds then stalls (console `$('video').currentTime`: 24.096208 and 24.080458 from 2 samples).
- I see a single buffered range 0, 36 on the video element in failure cases.

I see other stoppage issues, none quite as consistent as the above. Sometimes the video stops earlier, sometimes later. For release versions of Firefox without changing shaka player settings the above happens very reliably. I see plenty of issues with this demo in Chrome also. Not the same as above, but frequent stalls, failure to seek.

This looks to affect all released versions. ESR looks to have a widevine bustage at the moment which prevents me checking if it's affected. If I run Mozregression over this it looks like a regression however I wonder the following:

- We have components affecting our dynamic resolution
- We have a potential underlying issue which is being highlighted by a particular configuration of the above components
 
I'm uncertain as to if the fault is in Firefox or Shaka-player here (or some combination of the two).

Bisection data, take it with a large grain of salt, but incase it's useful.

Initial start date: 2017-09-05 (bad)
Initial end date: 2017-06-06 (good)
Regression within: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=fef489e8c2a193dde885adc48deb74cc883a5881&tochange=11755fd63168581e194258d04bb6a7337779ec78

Comment 3

[Bryce Seager van Dyk [:bryce] (he/him) - Not reading bugmail]

Ignore the above regression info. It looks to me more like this is intermittent, but will often manifest in the way detailed above. However, I'm seeing this is a lot of builds I had thought were good, particularly if you let the video play long enough.

Comment 4

[Bryce Seager van Dyk [:bryce] (he/him) - Not reading bugmail]

Not sure what just happened with those flags :\

Comment 5

[Bryce Seager van Dyk [:bryce] (he/him) - Not reading bugmail]

All right, we're gonna leave the flags alone for a bit until they stop being weird/busted.

Comment 6

[Chris Peterson [:cpeterson]]

Unassigning bugs assigned to Bryce because he no longer works at Mozilla.