We use DVR recording API to writing our incoming multicast channels per day. Recently I wrote that we can't get 24 hours of recording and we learned that the problem refers to garbage in our streams and wowza just skipping that packets. But then we switched to very clean source and we see that garbage just disappear. So we expected that we get our 24 hours of recording per day but not. Only ~22.
I've searched for a while for logs and see that our UDP incoming multicast streams restarting after timeout of 12 sec. And I think that this is key for our problem but I can't solve it.
restarting messages like:
INFO server comment - RTPMediaCaster.streamTimeout[1604382968:live/_definst_:kids_co_360p.stream]: timeout:12000 diff:17044 reason:101
INFO server comment - RTPMediaCaster.streamTimeout[1521938905:live/_definst_:expert_tv_270p.stream]: timeout:12000 diff:17054 reason:101
INFO server comment - RTPMediaCaster.streamTimeout[1437066670:live/_definst_:hd_travelch_360p.stream]: timeout:12000 diff:13539 reason:101
INFO server comment - RTPMediaCaster.streamTimeout[1450492277:live/_definst_:france24_576p.stream]: timeout:12000 diff:15566 reason:101
How do you think what's wrong?
Is it problem with our incoming streams or in wowza configuration?
I've started to use MediaCasterStreamMonitorAdvanced http://www.wowza.com/forums/content....onitorAdvanced with its default settings to see what's wrong and the problem is here:
But why wowza thinks that stream timeout??? I've monitored in next console by tcpdump that all packets continiously send to wowza.
2012-03-23 16:59:14 MSK comment server INFO 200 - ModuleMediaCasterStreamMonitorAdvanced.onValidateMediaCaster[live/_definst_] Stream not healthy [stream timeout]: flv:pervii_kanal.stream - - - 523.795 - - - - - - - - - - - - - - - - - - - - - - - - -
I've already tried to switch between wowza versions: Wowza Media Server 3.0.5.02 build1257 and latest Wowza 2. This problem exists.
I change option RTPTransportMode to UDP (also try with interleave):
And I try to dump our udp multicast mpeg-ts incoming streams with mplayer:
<!-- udp, interleave -->
and it dumps fine without any interruptions.
mplayer udp://184.108.40.206:10001 -v -dumpstream -dumpfile 1.ts
Here is a piece of our dumped stream:
Here is my debug screenlog. And confs and logs that tared:
So there is 1 problem that as I think due to another:
Problem is: dvr stores not all stream
Because: (as I think) wowza reconnects after timeout to our streams and then a piece of stream get lost.
Please help to solve this problem. Thanks.