GNOME Bugzilla – Bug 651031
Sometimes written file becomes corrupt when multicast stream stops
Last modified: 2011-05-25 18:00:13 UTC
Last night I have tested the writing to files from 5 cameras simultaneously. All of them were in mode "RTP over multicast". After 15 hours of testing I started to replace multicast group addresses from the camera's web interface (it was the stress test). The write process stopped. Then I obtained 'totem' player can normally play only 2 of 5 files. For other files it shows duration of 3 sec and seeking doesn't work. After launching 'ffmpeg' with key '-vcodec copy' new corrected files plays normally. This is the command line used for writing: 'gst-launch-0.10 -v rtspsrc location=rtsp://$1:7070 debug=true latency=0 protocols=0x02 ! rtpjpegdepay ! matroskamux ! filesink location=~/media/$1.mkv'
What exactly happens as the result of stopping the write process? If an EOS event is sent downstream you might need to wait until you receive the EOS message before stopping the pipeline, instead of stopping it when receiving the ERROR message. Could you get a debug log?
What happens is I change multicast group address setting in camera. According to recomendations of 'wtay' from irc I have added extra option '-e'. The bug didn't appear yet but it needs more testing.
That's probably a duplicate of bug #630997 *** This bug has been marked as a duplicate of bug 630997 ***