Connect to remote ffmpeg servers. Are you tired of unsuccessfully trying to pass your GPU through to a docker
container running in a VM? So was I! ffmpeg-over-ip allows you to run an ffmpeg server on a machine with access
to a GPU (Linux, Windows, or Mac) and connect to it from a remote machine. The only thing you need is Node.js
installed and a shared filesystem (could be NFS, SMB, etc.) between the two machines.
ffmpeg-over-ip consists of two main parts, the server and the client. Both are packed neatly into single JS
files. You can download these from the npm interface or by npm install ffmpeg-over-ip and then copying
them to the relevant places. You don't need any node_modules to run the server or the client.
The server and the client are both configured using JSONC (JSON with comments) configuration files. The paths
of these files can be flexible. To identify which paths are being used, you can invoke either with --debug-print-search-paths.
Template/example configuration files are provided in this repository for your convinience. Unless the server and the client
share the same filesystem, you may have to specify rewrites in the server configuration file.
Both the server and the client files are executable, so long as there is a Node.js installation available. If you intend
to use this in a docker container, you can directly mount the client file to where the container would expect a regular
ffmpeg executable to be, ie docker run -v ./path/to/client.js:/usr/lib/jellyfin-ffmpeg/ffmpeg ....
The server and the client communicate commands over HTTP, so make sure that whatever port you specify on the server is allowed through the firewall.
The contents of this project are licensed under the terms of the MIT License.