Jellyfind stopped recognizing my library after Linux update

[Daniel_King]

Hello,

I'm running Jellyfin on Linux with Portainer. My media is stored on Google Drive and I use Rclone to mount it.
This setup worked for quite some time, but after updating my Linux, my library won't be recognized any longer.

Some Portainer settings:

PGID: 1000
PUID: 1000


Host/volume  >>  Path in container 

/media/jellyfin/movies  >>  /data/Movies 
/media/jellyfin/shows  >>  /data/Shows


I mounted my corresponding Google Drive folders like this via Rclone:

nohup rclone mount gdrive:/ /media/jellyfin/movies  --allow-other  --buffer-size 64M  --dir-cache-time 72h  --drive-chunk-size 32M  --vfs-read-chunk-size 128M  --vfs-read-chunk-size-limit off  --vfs-cache-mode writes > rclone.log 2>&1 &

nohup rclone mount gdrive2:/ /media/jellyfin/shows  --allow-other  --buffer-size 64M  --dir-cache-time 72h  --drive-chunk-size 32M  --vfs-read-chunk-size 128M  --vfs-read-chunk-size-limit off  --vfs-cache-mode writes > rclone.log 2>&1 &


When performing a ls with movies and shows, I see my Google Drive content. So the mounting itself works.


Opening up Jellyfin, and trying to add a new library also shows me the /data/Movies and /data/Shows. Just when I perform a library scan or when I want to refresh the metadata, nothing is happening.


I have the feeling some permission must be off. Not sure if this is the reason, but when I check the permission with ls -l for my two mounted folders under /media/jellyfin I see this:

drwxr-xr-x 1 root root  0 Jan 24 10:59 movies
drwxr-xr-x 1 root root  0 Jan 24 10:59 shows

As my PGID & PUID are set to 1000 within Docker, could this mismatch be the reason?

Thanks in advance.

[TheDreadPirate]

Can you share your full Jellyfin logs via privatebin.net?

Since you running the container as UID 1000, that would correspond to your user. If your user can see the contents, the container should also have permissions to see the contents.

[Daniel_King]

Logs:
https://privatebin.net/?51ba6de4cf2e847e...C87ZKjC2zp

PW:
TheDreadPirate

[TheDreadPirate]

Your cloud storage is inaccessible MOST of the time. I see logs where it appears it can access it and then later it can't.

Does rclone generate its own logs that you could look at?

[Daniel_King]

I do, but it is more or less empty:

nohup: ignoring input


I need to check how to activate logs.
But how can I have troubles accessing the Drive if the mount works and I can see the files?

[TheDreadPirate]

I spoke with a jellyfin dev who uses rclone. You need to add --default-permissions to your rclone command. Otherwise rclone won't enforce linux permissions.

They also recommended you create a service instead of relying on nohup.

They provided an example service file.

Code:

# /etc/systemd/system/rclone.service
[Unit]
Description=Google Drive (rclone)
Requires=systemd-networkd.service
After=network-online.target
Wants=network-online.target
AssertPathIsDirectory=/media/cloud/

[Service]
Type=notify
KillMode=none
Environment=RCLONE_CONFIG=/opt/rclone/rclone.conf
RestartSec=5
ExecStart=/usr/bin/rclone mount gdrivecrypt: /media/cloud \
    --allow-other \
    --default-permissions \
    --uid 1005 \
    --gid 1007 \
    --dir-perms 777 \
    --file-perms 666 \
    --umask 002 \
    --log-file /opt/rclone/logs/rclone.log \
    --log-level NOTICE \
    --dir-cache-time 5000h \
    --drive-pacer-min-sleep 10ms \
    --drive-pacer-burst 200 \
    --poll-interval 30s \
    --cache-dir=/opt/rclone/cache \
    --drive-chunk-size 32M \
    --vfs-cache-mode full \
    --vfs-cache-max-size 350G \
    --vfs-cache-max-age 15m \
    --vfs-cache-poll-interval 5m \
    --vfs-read-ahead 2G \
    --vfs-read-chunk-size 32M \
    --vfs-read-chunk-size-limit off \
    --vfs-write-back 10m \
    --bwlimit 8650k:off \
    --transfers 3 \
    --rc \
    --rc-addr :5572 \
    --rc-web-gui \
    --rc-user USER \
    --rc-pass PASS \
    --bwlimit-file 20M
ExecStop=/bin/fusermount -uz /media/cloud
ExecStartPost=/usr/bin/rclone rc vfs/refresh recursive=true --rc-addr 127.0.0.1:5572 --rc-user=USER --rc-pass=PASS _async=true
Restart=on-failure
User=rclone
Group=rclone

[Install]
WantedBy=multi-user.target

[Daniel_King]

Thanks for the hint.

Well, as my setup used to work, I'm not sure if it is just because of the missing Rclone flag.
Nevertheless, I added it, it didn't change anything.

So I copied one test file directly to the server, and set up another library, just to see if Jellyfin can access the local one.

When I was afterwards scanning for the 3 library, out of the blue, my other two libraries started to sync something too.
But just one movie and one show got loaded, the test library also loaded the local file.
During the scanning process, my VPS got unresponsive and stopped to react. This also happened with second and third try.

Now I have the feeling that maybe the hardware (2 cores, 2 GB) of my VPS is not sufficient, and it just worked in the beginning because my library was not that big.


Does the service file solution also bring some performance enhancements?

[TheDreadPirate]

Yes and no. Yes, it would improve responsiveness on the remote side of the rclone setup as it optimizes caching. But this wouldn't overcome the limitations of your VPS.