RTSP stream stops after ~20 seconds

[MikeQBF]

Environment is networked video from appliance MPEG-4 encoder. No error codes or other log messages when stream stops. Problem is the same with 0.9.8 and 0.9.9 regardless of OS platform; on MacOS it will usually crash shortly after stopping, on Windows and Linux you can restart the stream, with the same end result.

What's really odd is there is no problem with 0.8.6 - works great, as a matter of fact, under "virgin" defaults. Same goes for QuickTime, no problems, keeps right on going (and going).

I'm baffled after spending most of the day hammering on codec settings. Any insights?

[Jean-Baptiste Kempf]

Messages?

[MikeQBF]

None - not a thing after the normal startup chatter. It just stops. A colleague confirmed that it also breaks standalone live555 the same way.

[Jean-Baptiste Kempf]

There are always messages...

[MikeQBF]

OK, j-b, I'll dig around yet again, but this is one of the issues that is leaving me scratching my head over this. I get nothing whether running through a GUI or directly from the command line with -vvv, etc. Makes it a pain to diagnose, yes?

[Rémi Denis-Courmont]

Make sure you're still receiving data. It sounds like a firewall or the server is timing out earlier than specified.

[MikeQBF]

Thanks, guys... we're getting closer.

j-b - yes, there are messages (MacOS crashing was interfering with message log, moved the testing to Ubuntu). All it's telling me is that the stream has stopped...

Rémi - ...and, yes, in monitoring the stream it has indeed stopped at the encoder. Thank you for the "forehead slap" lead on the problem.

At the moment we are pursuing a possible difference in the RTSP handshake between 0.8.6 and 0.9.x which might be changing (or setting) a timeout value in the stream server.

[MikeQBF]

We just confirmed that this is a problem with VLC and the use of GET_PARAMETER as an RTSP stream keep-alive. The latest versions don't "ping" the encoder/server with this keep-alive handshake. This was broken with 0.8.6e, and the problem documented with trac Ticket #1881. Unfortunately, it was recorded as "fixed" when it apparently isn't.

Do we reopen the ticket, or create a new one?

[Jean-Baptiste Kempf]

Create a new one.

[bipinskumar]

Is this issue fixed in the new version 1.0.3 ?