Is your feature request related to a problem? Please describe.
I'm exploring using KVS to stream video, but am nervous that I'll drop frames if the wifi goes down for anymore than a few minutes, since it seems like KVS only buffers to memory and not disk.
Describe the solution you'd like
I'd love for KVS to buffer frames that weren't successfully sent to disk, so it can then try to re-send those frames when the connection comes back. I know that this won't work forever (disk has a limit on storage too!) but given that disk sizes are 10-50x+ the size of memory, I'm hoping it gives me enough buffer so that the chance of something happening goes way down.
Describe alternatives you've considered
I've also thought about recording video files to disk and then uploading them async in the background, but I lose out on streaming the video real-time and also have to setup a mechanism for recording / background uploading / cleaning up the uploaded files / etc. Streaming feels like a cleaner / easier solution to start with.
Is your feature request related to a problem? Please describe. I'm exploring using KVS to stream video, but am nervous that I'll drop frames if the wifi goes down for anymore than a few minutes, since it seems like KVS only buffers to memory and not disk.
Describe the solution you'd like I'd love for KVS to buffer frames that weren't successfully sent to disk, so it can then try to re-send those frames when the connection comes back. I know that this won't work forever (disk has a limit on storage too!) but given that disk sizes are 10-50x+ the size of memory, I'm hoping it gives me enough buffer so that the chance of something happening goes way down.
Describe alternatives you've considered I've also thought about recording video files to disk and then uploading them async in the background, but I lose out on streaming the video real-time and also have to setup a mechanism for recording / background uploading / cleaning up the uploaded files / etc. Streaming feels like a cleaner / easier solution to start with.
Additional context It seems like there's been discussion of supporting this here from a few years ago, but I'm not sure if it ever ended up getting implemented.
If this isn't on the short term roadmap, would also appreciate any other ideas you may have. Thank you!