I have an ABR transcode scenario that is my master HLS output.
Prior to that, I have a a hot swap Failover setup in case stream goes down (looped file), and an Emergency hotswap so I can replace the stream.
This is a 24-hour stream, so in my scenario, the use case is:
Stream A (main incoming endpoint) - plays 20 hours per day when show not live
Stream B (emergency swap) - plays 4 hours per day when show is live
Fallback Loop (failover is both streams off)
All are transcoded to same resolutions/audio bitrates, etc. There is a 2-hour DVR cache.
This works just fine.
However, if we're sending RTMP to Stream B and the internet connection momentarily goes out, the audio will go out of sync with the video. Sometimes it's only only an issue on 1 or 2 renditions (doing 360, 720 and 1080), while one remains in sync, or sometimes they all go wacky.
The only way to fix (that I've been able to find) to stop nimble service, clear the dvr cache, and restart the service. Simply stopping and restarting incoming RTMP (originating from vMix) is not enough.
I don't know if this issue occurs if you're just doing a straight RTMP to ABR transcode (without all failovers we have), but it definitely happens and is repeatable in our setup.
Anything to look at or way to correct? We're 95% of the way to having everything work correctly all the time - this is kind of the last stumbling block.