Hi Favati,
I knew what your videos would should before even watching them, because I have the exact same problem here when using SDR# to connect via rtl_tcp to a Raspberry Pi.
Recent improvements to rtl_tcp have helped a lot with re-starting but this a new problem, as I'm accustomed to running the tcp connection for hours without such failures. I'm fairly certain the problem is on the sending (i.e. R-Pi) end, but haven't got data to prove this theory. However, since you can see in the video that SDR# continues to run, producing both audio and spectrum and waterfall displays, I think this shows that it is processing the IQ data which it is receiving (which is no longer correct).
Hopefully a fix will be forthcoming, as this is a pretty big problem for remote users. I'll certainly be happy to provide any data or help with evaluation.
73, Bob W9RAN
So...there are 2 possible problem (my english is not good at all!)
-the raspberry (maybe something relating to the tcp/ip)
or
-rtl_tcp bug
Or i've not understood you?
I'm investigating a lot...
I don't think it's a raspberry problem... when the problem occour (only in that moment!!), i can stress the raspberry with file transfer...the files are transfered ok.
I think it's a problem of rtl_tcp...
What kind of tests do you suggest now?
I've never seen that problem elsewhere (can you point me to other having this issue???????).