With all this talk about vfs cache preferences I wonder if rclone is scheduled to discontinue cache backend support before it's even out of beta?
I ask because I've noticed for my own testing purposes that friends say videos play faster and more seamlessly when i mount using cache backend rather than the numerous vfs cache configs i've tried.
At the end of the day i've found that either plexdrive5 or cache backend work best for me.
Since plexdrive is no longer being developed i'd rather stick with rclone mount, but I think it'd be a pitty to stop development on cache backend due to the overwhelming preference for vfs cache.
I'd like vfs cache to work for me.
Maybe it can if i knew exactly why plexdrive is so much better for performance for me. Maybe emulating what plexdrive does and translating that via vfs cache in my config/setup would yield better performance?
As @Animosity022 said, I don't currently have a maintainer for the cache backend. I'll keep it working as best as I can, but I've got too many other things to do to add new features to it at the moment.
In an ideal world the vfs cache would work for you . However the vs cache and backend cache have different aims - the vfs cache's main aim is to improve file system compatibility whereas the backend cache's aim is to make video playback smoother.
I great really great performance and playback with not using the cache backend. There are quite the number of use cases, the problem / challenge is that many people post as Plex is a beast of an ecosystem so making it work without rclone is often a challenge.
I find the streaming performance on the VFS alone to be quite good. Video opens in a couple of seconds, and never get any stutter at all even on high bitrate 4K. I would check if videos run ok outside of plex before assuming it's rclones fault. If they run fine outside plex but not inside then it's probably due to special considerations you need to have in the plex config. See Animositys recommendations on this (he has a post dedicated to his recommended setup including Plex).
What is really missing on the VFS currently is proper read-caching for the purpose of generally trading space for faster access to often used files. the VFS cache only has "full" mode which has horrendously bad drawbacks and I'd like to see a "readwrite" mode eventually that can cache partial read files while not forcing a full download and removing the ability to stream or have a pleasant user-experience. At that point I feel the old cache backend will probably be fairly redundant since it has a lot of other drawbacks that the VFS cache doesn't.
From a development perspective it also just seems wiser to add in the (relatively little) missing functionality to the VFS cache that is in active development by NCW than it is to do the much harder task of building upon unfamiliar code-base made by a developer who isn't active now and may not be coming back to continue the work.