[Bug 655804] New: Play position advancing but frames not rendered

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=655804
  GStreamer | gstreamer (core) | 0.10.34

           Summary: Play position advancing but frames not rendered
    Classification: Platform
           Product: GStreamer
           Version: 0.10.34
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: gstreamer (core)
        AssignedTo: gstreamer-bugs at lists.freedesktop.org
        ReportedBy: tvrtko.ursulin at onelan.co.uk
         QAContact: gstreamer-bugs at lists.freedesktop.org
     GNOME version: ---


Hi all,

When playing an MPEG TS stream multicast over UDP I occasionally see GStreamer 
stop rendering frames while the play position is normally progressing.

On the last occasion this happened it actually managed to untangle itself 
after a while and started playing (rendering) again. Unfortunately I do not 
know whether it would always be case since previously I did not wait this 
long.

To start getting more specific...

I do not know when exactly it stopped rendering, but I know when I first saw it 
in that state which was at 14:14 today. It was streaming for roughly two hours 
up to that point.

Last interesting logged thing before that state was a QOS event at 14:06. Then 
at 14:21 it started to render again and reported another QOS event. 

In between the application was querying play position every ten seconds which 
showed progress.

If you look at this annotated log:

... <video state unknown>
14:06:15.006 PLAY POSITION 6615867197374L
14:06:18.408 QOS @ 6619179855556L Frames: 291654/704; Jitter: 88265288
14:14 <spotted video not playing (not rendering)>
14:20:59.484 PLAY POSITION 7500344289211L
14:21:08.142 QOS @ 6915779855556L Frames: 306475/705; Jitter: 593222818788L
<video started playing again>

Play position is what query_position() on playbin2 element returns 
(documentation says this is stream time). And times logged in QOS events come 
from stream-time message field, so it should be matched up in that respect.

This second event is reporting that a frame was almost ten minutes late, but 
that there was only one dropped frame in the period on roughly a correct 
number of rendered frames (for 25 fps video) during the period. Which is of 
course wrong because nothing was rendered.

Application is written in python and uses playbin2 with a custom bin set as 
video-sink which is queue ! deinterlace ! queue ! xvimagesink. Audio-sink is 
set to a bin containing alsasink only.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.