Storage

clustercode recursively scans this folder for new video files that need encoding. But, don’t just mount your whole library here! clustercode supports prioritized tasks, meaning that you can control the order of the conversions. But to find out which order, you need to create subdirectories with the number as priority and put your videos in here.

Now if you don’t care about the order, just mount everything in /input/0/your-movie-lib-here. But here is another example, which might make sense.

Now, every TV episode will be picked up before all movies. If another cluster member is currently converting the last video in /input/1/, then clustercode will start looking in /input/0. The numbers do not need to be subsequent, but the higher it is, the higher the priority. Only use positive integers, so /input/0 is your least important input folder, ever! However the priority number does not make the conversion faster in any way. Also, note thate the input directories need write permissions.

Ideally every cluster member has access to the same library, meaning that it’s a shared storage (NFS, GlusterFS etc). If you have a cluster member that has additional or different video files to convert, mount them in another priority.

Of course, only Member B will encode the anime files, as A doesn’t have access to. The way clustercode checks if another task is currently being processed by a node is to compare the relative file paths, e.g. 0/movies/a-movie.mkv. With different priorities it is ensured that the same video is not being processed multiple times.

The converted video files will be stored here. But only after they have been successfully encoded. There are strategies how the cleanup exactly happens and which directory structure within is being applied. See cleanup strategies for more details.

The profiles are stored here. At first run, default.ffmpeg and x265.ffmpeg will be created here. The default uses common x264 parameters for ffmpeg and is the last resort for profiles if others couldn’t be matched.

This is where the currently processing video file is being created. In Docker containers the files are stored with the image filesystem by default, rather than a tmpfs in /tmp, which consumes RAM. If you have special storage needs, you can mount it from outside. If a container stops during conversion, this directory will not get cleaned and the conversion process cannot resume from where it left.