Hi,

I’ve been having a recurring issue with my Unraid server and I’m trying to understand what’s going on.

From time to time, the web UI becomes unusable. I can reach the login page, but when I enter my credentials it just refreshes in a loop and never logs in.

At the same time, Docker containers are still running. Plex is accessible, I can browse the library, but playback fails because the transcoder stops responding. It throws a playback/transcoder error.

A reboot fixes everything, but the problem eventually comes back.

It feels like something is getting saturated or locked up. The system isn’t fully dead, just “half alive”.

For context:

• Custom Unraid build

• i5-13500

• 64GB RAM

• Multiple SSDs + HDD array

• Plex in Docker

• Transcoding enabled

I’m wondering if this could be:

• RAM getting full

• /tmp filling up

• Docker image corruption

• Logs filling up

• Some nginx or GUI process crashing

• OOM killer killing something important

Has anyone seen similar behavior?
What should I check next time it happens before rebooting?

Thanks.

tower-diagnostics-20260211-1734.zip

18 minutes ago, Pinedours said:

What should I check next time it happens before rebooting?

Get diagnostics if you can

Also

df -h /

After a fresh reboot, everything looks normal, but a few hours and a some plex transcodes after , "df -h /" command tell me that my rootfs is at about 91%... it is the case actually and this is there is my diagnostic files attached to this message..
thx for your help.

tower-diagnostics-20260213-1622.zip

I have the impression, that it spikes specifically during mobile downloads too....

The usual cause of filling rootfs is something writing to a path that isn't actual mounted storage. Often this will be some host path that isn't a user share, disk or pool.

18 minutes ago, Pinedours said:

some plex transcodes

Post docker run for plex.

docker run
  -d
  --name='plex'
  --net='host'
  --pids-limit 2048
  -e TZ="Europe/Paris"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Tower"
  -e HOST_CONTAINERNAME="plex"
  -e 'VERSION'='docker'
  -e 'PLEX_CLAIM'='XXXXXXXXXXXXXXXXX'
  -e 'PUID'='99'
  -e 'PGID'='100'
  -e 'UMASK'='022'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:32400]/web/index.html'
  -l net.unraid.docker.icon=''
  -v '/mnt/unionfs/Media':'/data':'ro'
  -v '/mnt/user/transcodes':'/transcode':'rw'
  -v '/mnt/user/appdata/plex':'/config':'rw'
  --device='/dev/dri' 'lscr.io/linuxserver/plex'

3 minutes ago, Pinedours said:

-v '/mnt/unionfs/Media':'/data':'ro'

Explain this

"/mnt/unionfs/Media" is not a simple disk path.

It is a mergerfs mount that combines:

• a local storage path ("/mnt/user/local/Media")

• a remote rclone mount ("/mnt/remotes/dropbox-crypt/Media")

So effectively, Plex is reading from a merged filesystem that includes both:

• local media

• encrypted Dropbox remote storage

Edited February 13Feb 13 by Pinedours

I would think that should at least be ro:slave.

Do you still have the problem if you eliminate that and just use the user share?

Just to pipe in, this might be 2 seperate issues.

I found this issue as my GUI was stuck on the login page, it would accept the correct credentials but just refresh the page. Wrong credentials would show an errror message.

I could however still access unraid using a different browser, so i cleared cookies and it worked.

There might be an issue with unraid interacting with firefox if you're using it to get to the GUI too.

Edited February 14Feb 14 by ThatDeltaGuy