I use the mediatomb with encoding option to capture live stream in FLV, which is then encoded by mencoder to MPEG2 and served. The problem is that there some time required (3-5 secs usually) to appear first data in the buffer. If I try, then, to open resource from my UPNP client, I see mutiple times that mediatomb creates the fifo, waits some time and then kills the buffering thread, closes the fifo and creates new fifo+transcodind thread. This may be a never-ending store and I may never see the content of video because of constant killing/spawning new transcoding thread.
Few question about that:
* what's the rule when the mediatomb decides to kill the thread and spawn new one, does it depend on UPNP client implementation? (ie. might it be so unpatient that closes the request and creates new one if there's no data since few seconds?)
* can I configure this behaviour somewhere or patch the code to avoid this?