My VFS SweetSpot - Updated 30-Aug-2018


Thanks for the reply! I switched to vfs only and it looks to be running perfectly smoothly. I thought the cache backend would really help out but I suppose it didn’t.



Also remember how cache / vfs expiration works. Let’s say you have a folder called “Movies” and you add a new movie to that folder, let’s say Movies\Solo (2018)

Cache/VFS will expire your Movies folder and will need to rescan the whole thing. So if you have 2000 movies it’ll take a while for rebuild.

I’m not sure if this is a google limitation or what but it sucks ;(



That’s probably what’s happening. If I haven’t ran a time on the remote, where all things are downloaded and changes occur, in a couple of hours then the command takes ~10 minutes to scan through again. If you can confirm that’s the behavior, and adding anything to a third-level subfolder means the top-level has to be refreshed and scanned in for quick access, then that may be the issue I’ve been trying to resolve. Makes those Kodi scans even longer…


@Animosity022 checked your script on GitHub, seen you added a buffer-size of 1GB. Is this working ok now since the latest release of rclone? I’ve read in the thread there were some problems and using the standard buffer size was better.


No issues at all with 1GB. I’ve been pretty rock solid since I figured out the ATV Direct Play issue.


Cool, i’ll add it back in my version. The “Direct play” issue is only on the AppleTV?


I only have ATVs in my house and it would occur on the 4K version and the regular version.

I was going to test again once TVOS 12 hits but I have been using a reverse proxy for plex (Caddy) and that seems to make the issue go away for me as it doesn’t open/close the files repeatedly anymore.


Ah curious if tvOS 12 will fix it, any reason you chose Caddy over Nginx or treaffik? Already familiar or The choice was specific for that bug?


I’ve used Nginx for quite a bit. I picked Caddy as the config is insanely simple compared to NGinx.

I never heard of Treaffik but I did search and will check it out as it does seem like something to check out.

I’d think any proxy would mitigate the issue.

TVOS12 doesn’t release till Monday so I haven’t tried it yet.


Ah ok, all my plex servers are behind nginx proxy. Haven’t looked into Caddy since I’ve switched from Apache (macOS) to nginx. Traefik looks to have nice features, specially when using docker as it can auto adjust itself depending on what’s running behind it. When I got time i’ll look into it also.

Monday it’s d day for the non dev Apple people? :wink:


Caddy easily plugs into Google OAuth, which was a big thing for me as I can authenticate into my internal apps via my Google Account, which is super cool and don’t have to worry about form logins or basic password auth.

I had find a 3rd party plugin with NGinx to do that, which did work without an issue as well.


@Animosity022 how long on average does your media take to start after you hit play?


Usually 1-2 seconds at most.


I can’t really get faster then 10 seconds. I can live with it but I would love to get it faster one day.


And you direct playing? ( no transcoding)


same here from hetzner dedi to uk isp


Yep. Direct play, full quality. Whatever you want to call it.


Direct Play isn’t the same as full quality.

Direct Play is when it doesn’t have to transcode anything.
Direct Stream is a very light CPU transcode.
Transcoding is when it has do perform a bit more work.

You’d see this in the status screen as to which of the 3 it is doing.


Isn’t it though? Direct Play just plays the source file without any conversion (except that done by the player you use for Plex). It might be better to say source quality rather than full quality if its a re-encoded file of the original.


They are two different settings in Plex.

You can turn on Direct Play via the “Allow Direct Play” setting which is On/Off.

Quality is another setting, which in turn can impact something direct playing or not as if you have quality turned down, you’ll transcode and not direct play.