DZMM

With only 4GB you might need to go even smaller if you've got concurrent usage as I think the buffer is per stream. I see you also removed --vfs-read-chunk-size-limit off from your mount script. This defaults to off so removing won't help - if you're still having memory problems I'd set this again to something like 256MB of 512MB, but you're going to struggle to play remux files IMHO if you don't have enough ram to buffer the stream. I don't know what size chunks my setup requests, but I have no issues. If you're happy with only being able to play files once they've been uploaded then this is fine. However, if you have a lot of files that are pending upload and you want to Plex/Sonarr/Radarr etc to be able to see them, then this is where the unionfs and rclone_upload folders come into play - e.g. unless you make your setup a lot more complicated you can only upload 750GB/day. There are ways around this e.g. my setup allows me to upload at my full 1G line speed 247, but it took me a few hours to setup, but it means I don't have files waiting to be uploaded for more than a few mins. I still need the unionfs setup though so that Plex doesn't have to rescan thinking the file has changed when it moves, when it hasn't etc etc Check out the new scripts on github as it accounts for this e.g. if it tries to mount and there's no internet connection then it tries again a few mins later. This check is useful if your mount ever drops as it will automatically remount.

just stop radarr, sonarr, plex etc - anything that's accessing the mount before you unmount. Shouldn't have any problems

hmm have you tried unmounting and removing from rclone_upload or directly from the mount?

remove manually?

pfsense traffic shaping does shape downstream - it does this by dropping packets, so that's why setting your bandwidth settings is important.

But from the official documentation I can see Why have you enabled the Quick flag? Ignore my first version as it didn't work properly. It should do. I use floating rules as it was easier for me to have all in one place, and also you can adjust the order then.

I'm 99% certain that at least for Plex no disks are used to play files - everything is streamed using RAM. I think even the rclone dir_cache is stored in RAM. I'm not sure how Kodi works i.e. is the file downloaded before playing - I don't think so. The thread has kind of got long, which is good - at first I was worried no-one was interested but the collaboration has helped improve setups, including mine. I've never had an API ban and my library is reasonably big - there are users in this thread with libraries x5 the size of mine and over a Petabyte. I haven't had any problems or even stuttering for a couple of months, and all my content is in a team drive now: root@Highlander:~# rclone size tdrive_vfs: Total objects: 53947 Total size: 248.142 TBytes (272835090178653 Bytes)

Use Screen from nerd pack

You should avoid trying to write direct to the mount i.e. to /mount_rclone - always try to write to /mount_unionfs where the file(s) will get stored for a future upload in the background via /rclone_upload. You can also add files direct to /rclone_upload if you want, but it's best to use /mount_unionfs unless you know what you are doing so that you don't mess up sonarr, radarr etc Now to confuse matters - for files that are on gdrive that you want to move within the mount, I find the best way to move them is to move them in /mount_rclone rather than /mount_unionfs as this won't require you to download them just to re-upload, which moving within /mount_unionfs can result in.

Thanks - download worked. will reboot in the morning

no - you just need to run rclone config again on the remote

My guess is you haven't created your own client_id: https://rclone.org/drive/#making-your-own-client-id

Thanks. I got an error though: Now installing TBS (Open Source) version 6.7.0.rc6 Base URL: https://lsio.ams3.digitaloceanspaces.com/unraid-dvb/6-7-0-rc6/tbs-os Downloading: /tmp/mediabuild/bzimage ... done Downloading: /tmp/mediabuild/bzroot ... failed (Invalid URL / Server error response) Downloading: /tmp/mediabuild/bzroot-gui ... failed (Invalid URL / Server error response) Downloading: /tmp/mediabuild/bzmodules ... done Downloading: /tmp/mediabuild/bzfirmware ... done Downloading: /tmp/mediabuild/bzimage.md5 ... done Downloading: /tmp/mediabuild/bzroot.md5 ... done Downloading: /tmp/mediabuild/bzroot-gui.md5 ... done Downloading: /tmp/mediabuild/bzmodules.md5 ... done Downloading: /tmp/mediabuild/bzfirmware.md5 ... done Checking MD5's: failed!

If you use the settings in my guide no files get written to disk

If you're just mounting and not uploading files it all happens in memory. If you're uploading, then set your rclone_upload directory to cache only

I think your Plex library needed a rescan

That's a Plex error. From a quick glance of the Plex forums it looks like you need to rescan/check your library folder setup

@SoloLab All ok now?

Hmm the upload is independent of the mount so should work. If you run rclone config at the command prompt does it work? What's on line 25 of your script? The error message is one I added to the script, so isn't a 'real' one

36TB with about 25% utilisation - most of it is a big 'cache' for files waiting to be uploaded to Google - about 300TB there

In the rclone mount

Add --min-age 7d to your rclone upload command

If you're confident you won't be changing files eg using sonarr/radarr then yes you could use RW for everything. If not, RW on the mount would cause too many potential problems IMO

Do you use tdrives? I can't get my space used anymore. I don't have a PB like you though - insane! I think I'm around 250-350TB. I'd love to ditch unionfs as well for better delete support and because it doesn't support hardlinks. Mergerfs does, but I'm waiting for rclone union to be beefed up for a all-in-one solution, and because mergerfs isn't available on unraid.

I'm glad it worked, but you really need to have your Plex thumbnails on a SSD for speed!