[Live-devel] How to distinguish new RTP packets fromout-of-data RTP packets when random accessing the video

[Ross Finlayson]

> Are you actually seeing this happen with this server (a Darwin Streaming Server)?    Our RTP reception software will automatically discard out-of-order incoming RTP packets (by checking the RTP sequence number).  So, if you're seeing "old RTP packets" after the "PLAY", then presumably you're seeing some 'old RTP packets', followed by only 'new RTP packets'.  You should not be seeing 'old RTP packets' mixed with 'new RTP packets'.
>  
>  
> Yes. It happened. Sometimes, network traffic may be congested.

If you're seeing 'old RTP packets' mixed with 'new RTP packets' (i.e, not just 'old RTP packets", followed by 'new RTP packets'), then that means that much more than "network congestion" is happening here.  If that is what you're seeing, then packets are being not just delayed, but reordered - *excessively* reordered.  In fact, they're being reordered so much that our RTP sequence number check is not catching it - which means that some packets are out-of-order more than 32767 (==2^15) packets *within the same stream*.  Something is badly broken with your server (most likely), or your network.  That's what you should be fixing.


> So, we need more information (e.g. RTP header, command response…) to overcome the problem resulting from congested network.

Our RTP implementation already takes care of this - so that it should be delivering RTP payloads in order.  If that's not happening, then - as I noted above - your server or network is badly broken, and there's nothing that our software can do to overcome this.

The first thing I recommend that you do is check whether you have the latest software version for your "Darwin Streaming Server" (either that, or use a better server).

Ross Finlayson
Live Networks, Inc.
http://www.live555.com/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.live555.com/pipermail/live-devel/attachments/20120107/49834e3d/attachment-0001.html>