1 hour ago, trey81 said:

I fixed my issue with containers not starting/not using GPU after updating Unraid to 7.2.5 and hopefully this can help others as well. It was sated in other posts, but I didn't understand at the time.

Did you do a reboot also in between?

1 hour ago, trey81 said:

The container management switched to CDI (Container Device Interface). Which, what I've read, is a standardized way to expose hardware (GPUs... etc.) to containers and such. Looks like this update switched to this. Here's what I did to fix it:

Please try it without that "fix" since this is already in the plugin.

1 hour ago, trey81 said:

Enter in the NEW CDI format for your GPU into the 'Extra Parameters' field:

It should be actually okay to use the old way described in post 2 of this thread, or also --gpus all

1 hour ago, trey81 said:

--device=nvidia.com/gpu=all

1 hour ago, trey81 said:

After this, my containers worked and used my GPU again. I'm no expert and I cannot help beyond what I've typed here. The internet helped me understand this more. Hopefully you can get yours working, too.

If possible please try again after a reboot, since this seems a bit overcomplicated and it should now work right OOB.

4 hours ago, ich777 said:

Sorry about that, you won't see those message anymore with the plugin update 2026.05.07, this was displayed by accident.

Errors on bootup are gone - thank you

4 hours ago, ich777 said:

Where did you test that? In the browser or on a native app?

Tested with iOS app and also Nvidia Shield

4 hours ago, ich777 said:

Does the card show up in your Plex transcoding settings?

Yes I do see it in my plex settings under transcoding

I still don't see any processes found when i run smi on the plex docker.

Will test more later today and report back

Edited May 7May 7 by Viperkc

6 minutes ago, Viperkc said:

Errors on bootup are gone - thank you

These where no errors, just warnings, nothing to worry about.

7 minutes ago, Viperkc said:

I still don't see any processes found when i run smi on the plex docker.

Your docker run command?

7 minutes ago, Viperkc said:

I still don't see any processes found when i run smi on the plex docker.

Do you see the card in your Plex transcoding settings?

8 hours ago, ich777 said:

Your docker run command?

I have --runtime=nvidia under extra parameters

NVIDIA_DRIVER_CAPABILITIES = All

NVIDIA_VISIBLE_DEVICES - my device ID

Have been running this for 4+ yrs no issues

8 hours ago, ich777 said:

Do you see the card in your Plex transcoding settings?

Yes I do see the card in Plex settings

I just noticed that in the GPU plugin it does show processes and some transcoding - so maybe the only thing that is broken is the SMI just not showing the processes

I just had noticed high CPU usage once upgrading to 7.2.5 and just panicked thinking transcoding was not being done by my GPU and the CPU was doing it.

I know there are processes ATM.

Thanks for all your help on this. Now to figure out why my CPU usage is so high with 7.2.5 - pretty sure its not related to the Nvidia driver.

2 hours ago, Viperkc said:

I just had noticed high CPU usage once upgrading to 7.2.5 and just panicked thinking transcoding was not being done by my GPU and the CPU was doing it.

Probably backed in subtitles, these are a nightmare for your CPU.

2 hours ago, Viperkc said:

I just noticed that in the GPU plugin it does show processes and some transcoding - so maybe the only thing that is broken is the SMI just not showing the processes

Transcoding is actually Throttled by Plex, meaning it transcodes a batch, then ends the process, transcodes a batch and so on... in between you won't see any processes.

New plugin update is out. Does anyone know the proper steps to get it working again? Is it update then restart Unraid?

14 minutes ago, Nanuk_ said:

New plugin update is out. Does anyone know the proper steps to get it working again? Is it update then restart Unraid?

I don't understand.

1. Go to the Plugin page

2. Click Check for Updates

3. At the Nvidia Plugin click Update

4. Done

Nothing more needed.

However this update was a really small change which only changed a console output on boot.

If you have issues, please post your Diagnostics and what's not working exactly.

I'm trying to figure out if this is the right solution for me. I'm hitting a sw_power_cap even under ~50% load on my RTX 4000 running Ollama.

What would be right solution to fix this?

Running on unRAID 6.12.15 with the latest Nvidia driver 575.51.02

(I'm super new to managing my hardware down to this level so bare with me please)

On 4/22/2026 at 9:33 AM, fka said:

bin/bash

if [ "$2" == "prepare" ]
then
killall nvidia-persistenced
elif [ "$2" == "release" ]
then
nvidia-persistenced

echo auto | tee /sys/bus/pci/devices/????:??:??.?/power/control

fi

exit 0 

Edited May 8May 8 by d_buster

1 hour ago, d_buster said:

What would be right solution to fix this?

This is not an issue per se, this is just a power limitation from your card to not get damaged. However since it seems you are using Ollama this question would be best suited in the support thread from the container.

However as said, that's just normal and nothing to worry about.

The script that you've linked is for VMs and not needed when running applications in containers.

That makes sense, I appreciate the feedback.

The original reason why I ask is because the sw_power_cap triggers around 98-100 watt, the RTX4000 has a 160 watt TDP rating, I guess I expected throttling to kick on closer to 150 watt.

Updated without issue. Thanks guys!

I'm on a GTX 1050ti and no matter what I do, I can't get the plugin to download and install the v580.159.03 driver. I keep getting "Can't download Nvidia Driver Package v580.159.03". I don't know how to manually force it to install. I have tried to wget the file directly into the directory it looks in and it will verify the checksum and pass but still says it can't download (it even deletes the file for some reason). Running unRAID 7.2.6 with plugin version 2026.05.07

30 minutes ago, GoldenCyn said:

I'm on a GTX 1050ti and no matter what I do, I can't get the plugin to download and install the v580.159.03 driver. I keep getting "Can't download Nvidia Driver Package v580.159.03". I don't know how to manually force it to install. I have tried to wget the file directly into the directory it looks in and it will verify the checksum and pass but still says it can't download (it even deletes the file for some reason). Running unRAID 7.2.6 with plugin version 2026.05.07

How does the plugin page look like, do you have some php errors on there.

If you have php errros on the page please follow this procedure:

1. Remove the plugin

2. Reboot

3. Install the plugin again

4. Go to the plugin settings and select the 580.x driver and click Update & Download

5. Reboot

Diagnostics would also always help.

I've just update to 7.2.6 and got isseu with my plex container and nvidia p620. The container don't start anymore i had to remove the nvidia to let it start. I've tryesd to reinstall the driver ando also the aother trick indicate here but with not siccess anyone can help me?

Here is my error:

docker run
  -d
  --name='plex'
  --net='host'
  --pids-limit 2048
  -e TZ="Europe/Berlin"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Joshua"
  -e HOST_CONTAINERNAME="plex"
  -e 'VERSION'='public'
  -e 'PLEX_CLAIM'=''
  -e 'NVIDIA_VISIBLE_DEVICES'='GPU-1ec4894a-9a88-7038-922c-04c92d1742f6'
  -e 'NVIDIA_DRIVER_CAPABILITIES'='all'
  -e 'PUID'='99'
  -e 'PGID'='100'
  -e 'UMASK'='022'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:32400]'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/linuxserver/docker-templates/master/linuxserver.io/img/plex-logo.png'
  -v '/mnt/user/Show/':'/tv':'ro,slave'
  -v '/mnt/user/Movies/':'/movies':'ro,slave'
  -v '/tmp/':'/transcode':'rw'
  -v '/mnt/user/Video/':'/video':'ro,shared'
  -v '/mnt/user/appdata/plex':'/config':'rw'
  --runtime=nvidia 'lscr.io/linuxserver/plex'
42bca44bc77ea11a4fa5305ad54e5ba4bb88d753cd520f232f482b6511fce6e6
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error jailing process inside rootfs: open /proc/self/mountinfo: no such file or directory

Run 'docker run --help' for more information

The command failed.

Edited May 10May 10 by Alex_75

7 hours ago, Alex_75 said:

Here is my error:

Please post your Diagnostics.

i notice my zfs mirror get data corruption on both drive same file do mem test pass decide to downgrade from 7.2.6 to 7.2.5 the same thing happened lucky i have snapshot and backup problem solve when i downgrade from nvidia driver v595.71.05 to v580.159.03

bug product

• unraid 7.2.5-6

• zfs mirror with dataset

• nvidia driver v595.71.05

• RTX 4060

• copy new file or do zfs scub will report new data corruption

it's a silence data corruption no log no drive error only when i check zpool status -v cache

is this PCIe/ZFS I/O corruption bug?

Edited May 11May 11 by Gvon

2 hours ago, Gvon said:

it's a silence data corruption no log no drive error only when i check zpool status -v cache

is this PCIe/ZFS I/O corruption bug?

Sorry but I can't help with that since I don't think the Nvidia Driver is able to cause issues with ZFS and this is of course the first time I hear of such a thing.

You only can research or report that in the ZFS GitHub issue tracker or the Nvidia developer forums.

11 hours ago, ich777 said:

Please post your Diagnostics.

joshua-diagnostics-20260511-2030.zip

Here is the file

Thanks

8 minutes ago, Alex_75 said:

Here is the file

I'm not 100% sure but please try to remove --runtime nvidia and instead add --gpus all from the Extra parameters.

33 minutes ago, ich777 said:

Sorry but I can't help with that since I don't think the Nvidia Driver is able to cause issues with ZFS and this is of course the first time I hear of such a thing.

You only can research or report that in the ZFS GitHub issue tracker or the Nvidia developer forums.

this is what i found maybe related ?

Random screen freezes on Linux with driver 595.71.05 where even the keyboard stops responding — that's effectively a kernel freeze, which could block ZFS I/O mid-write.
https://forums.developer.nvidia.com/t/595-71-05-screen-freezes-on-linux-wayland/369122

Graphical corruption after suspend/resume with Xid 13 errors on 595.71.05
https://forums.developer.nvidia.com/t/graphical-corruption-after-suspend-resume-with-nvidia-xid-13-595-71-05/368627

54 minutes ago, Gvon said:

this is what i found maybe related ?

Maybe but I can't tell, it would be best to create your own Forum entry there.

55 minutes ago, Gvon said:

where even the keyboard stops responding — that's effectively a kernel freeze

Are you sure, who says that? I'm not sure if the keyboard really hangs because he mentions that he thinks that it's hanging. However from what I see this is an Xid error 79 (GPU falls from the Bus - most of the times a incompatible BIOS setting or even outdated BIOS.

55 minutes ago, Gvon said:

Xid 13 errors

I don't think that you have an Xid error or do you?

https://docs.nvidia.com/deploy/xid-errors/analyzing-xid-catalog.html

1 hour ago, ich777 said:

I'm not 100% sure but please try to remove --runtime nvidia and instead add --gpus all from the Extra parameters.

I've troubleshooted with chatgpt. I don't know what happened with the upgrade to 7.2.6 but i delete my docker image and reinstalled it and now all is working well

1 hour ago, Alex_75 said:

I've troubleshooted with chatgpt.

Most of the times a really bad choice if we are talking about Unraid and 3rd party drivers.

However glad that all now is working.

8 hours ago, ich777 said:

I'm not 100% sure but please try to remove --runtime nvidia and instead add --gpus all from the Extra parameters.

This fixed the issue for me in Plex after upgrading to Nvidia Driver Version: 580.159.03. I was getting this error when starting Plex:

docker: Error response from daemon: unknown or invalid runtime name: nvidia

Edited May 12May 12 by Naratov

2 hours ago, Naratov said:

This fixed the issue for me in Plex after upgrading to Nvidia Driver Version: 580.159.03. I was getting this error when starting Plex:

I don't think that this fixed it, is it possible that you've installed the Nvidia driver and didn't restart the Docker service or did not reboot?

The message that you are seeing means exactly that.