Yeah but the issue is that it automatically installs the latest driver and by downgrading I'm forced to reboot 😅

Anyway... It works now. Thank you everyone

Edited May 1May 1 by Ikeasofa

1 hour ago, Ikeasofa said:

Yeah but the issue is that it automatically installs the latest driver and by downgrading I'm forced to reboot 😅

Anyway... It works now. Thank you everyone

Maybe time to buy a new gpu. 😜

With these prices 😭

1 minute ago, Ikeasofa said:

With these prices 😭

Would be funny if Nvidia said, yeah 580.x is now also deprecated. 🤪

But that won't happen, at least I think.

Update / findings concerning "failed to acquire modesetting permission issue" in the context of steam-headless and consecutive xorg, xvnc and sunshine crashloop:

After further testing and updating the plugin/driver to latest open source, the original "Failed to acquire modesetting permission" issue appears to be related to the NVIDIA 595 driver branch and the host GUI / DRM modesetting state, not simply a Docker or container configuration mistake.

before:
NVIDIA(GPU-0): Failed to acquire modesetting permission
xorg FATAL
Sunshine Found [0] outputs

after:
xorg RUNNING
DISPLAY :55 OK
Sunshine Found [1] outputs
Detected display: HDMI-0 connected: true
Found H.264 encoder: h264_nvenc
Found HEVC encoder: hevc_nvenc
Found AV1 encoder: av1_nvenc

There is now a separate second issue in the container stack: a crash/restart loop related to Sunshine virtual input devices and /dev/input/event* handling. That appears to be separate from the original NVIDIA modesetting problem. And ich77, you already wondered about the many "input devices" related lines with in my syslog, so another independent issue I try to adress at the steam-headless support forum and github respectively. Thank you so much for your work!

Edited May 1May 1 by anticholium

For clarity reasons i can say that the new version works as expected with v7.2.4. Update driver package and restarted server, all working, no issues with install or update..

Note I have not updated to v7.2.5 as of yet

6 hours ago, Ikeasofa said:

Yeah but the issue is that it automatically installs the latest driver and by downgrading I'm forced to reboot 😅

Anyway... It works now. Thank you everyone

I had the same issue with my P2000. Plugin has a bug where on upgrading the UnRaid from 7.2.4 to 7.2.5 it installed the latest driver that is incompatible with my GPU even though I had manually selected the 580 driver branch. Had to reboot the server 2 times after fixing this. Now it is okay though.

Cheers!

Despite updating to the latest plugin and swapping the driver, I'm still having issues.

Troubleshooting tried so far:

• Update plugin and reboot

• Upgrade driver (to 595) and reboot

• Uninstall driver, reboot, install driver, reboot

• Downgrade driver (to 580) and reboot
  

  I'm using an RTX 2000 Ada. The GPU ID is properly configured in my Plex and Tdarr dockers, but the GPU isn't able to be used.
  
  root@Storage:~# dmesg | grep -i nvidia

  [ 31.018588] nvidia: loading out-of-tree module taints kernel.

  [ 31.019314] nvidia: module license 'NVIDIA' taints kernel.

  [ 31.020648] nvidia: module license taints kernel.

  [ 31.231377] nvidia-nvlink: Nvlink Core is being initialized, major device number 239

  [ 31.240228] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem

  [ 31.289817] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.159.03 Fri Apr 24 06:16:47 UTC 2026

  [ 31.420257] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 580.159.03 Fri Apr 24 05:49:32 UTC 2026

  [ 31.423016] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver

  [ 31.424264] [drm] Initialized nvidia-drm 0.0.0 for 0000:02:00.0 on minor 1

  [ 83.268922] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.

The GPU is visible to the container:

root@Storage:~# docker exec -it Plex-Media-Server nvidia-smi

Sat May 2 06:27:18 2026

+-----------------------------------------------------------------------------------------+

| NVIDIA-SMI 580.159.03 Driver Version: 580.159.03 CUDA Version: 13.0 |

+-----------------------------------------+------------------------+----------------------+

| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |

| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |

| | | MIG M. |

|=========================================+========================+======================|

| 0 NVIDIA RTX 2000 Ada Gene... Off | 00000000:02:00.0 Off | Off |

| 30% 48C P0 17W / 70W | 0MiB / 16380MiB | 4% Default |

| | | N/A |

+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+

| Processes: |

| GPU GI CI PID Type Process name GPU Memory |

| ID ID Usage |

|=========================================================================================|

| No running processes found |

+-----------------------------------------------------------------------------------------+

Is there something obvious that I'm missing that would have been thrown out of whack with the latest unraid and driver updates? Anyone else having this experience?

2 hours ago, Bluephoria said:

Is there something obvious that I'm missing that would have been thrown out of whack with the latest unraid and driver updates? Anyone else having this experience?

No, everything seems fine.

With your RTX card, you should definitely use the latest driver.

6 hours ago, Bluephoria said:

Despite updating to the latest plugin and swapping the driver, I'm still having issues.

Troubleshooting tried so far:

• Update plugin and reboot

• Upgrade driver (to 595) and reboot

• Uninstall driver, reboot, install driver, reboot

• Downgrade driver (to 580) and reboot
  

  I'm using an RTX 2000 Ada. The GPU ID is properly configured in my Plex and Tdarr dockers, but the GPU isn't able to be used.
  
  root@Storage:~# dmesg | grep -i nvidia

  [ 31.018588] nvidia: loading out-of-tree module taints kernel.

  [ 31.019314] nvidia: module license 'NVIDIA' taints kernel.

  [ 31.020648] nvidia: module license taints kernel.

  [ 31.231377] nvidia-nvlink: Nvlink Core is being initialized, major device number 239

  [ 31.240228] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem

  [ 31.289817] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.159.03 Fri Apr 24 06:16:47 UTC 2026

  [ 31.420257] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 580.159.03 Fri Apr 24 05:49:32 UTC 2026

  [ 31.423016] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver

  [ 31.424264] [drm] Initialized nvidia-drm 0.0.0 for 0000:02:00.0 on minor 1

  [ 83.268922] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.

The GPU is visible to the container:

root@Storage:~# docker exec -it Plex-Media-Server nvidia-smi

Sat May 2 06:27:18 2026

+-----------------------------------------------------------------------------------------+

| NVIDIA-SMI 580.159.03 Driver Version: 580.159.03 CUDA Version: 13.0 |

+-----------------------------------------+------------------------+----------------------+

| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |

| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |

| | | MIG M. |

|=========================================+========================+======================|

| 0 NVIDIA RTX 2000 Ada Gene... Off | 00000000:02:00.0 Off | Off |

| 30% 48C P0 17W / 70W | 0MiB / 16380MiB | 4% Default |

| | | N/A |

+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+

| Processes: |

| GPU GI CI PID Type Process name GPU Memory |

| ID ID Usage |

|=========================================================================================|

| No running processes found |

+-----------------------------------------------------------------------------------------+

Is there something obvious that I'm missing that would have been thrown out of whack with the latest unraid and driver updates? Anyone else having this experience?

I had the same issue with a 3060. I followed the following steps from a reddit comment which got the gpu to work:

Same boat — Unraid 7.2.5, Nvidia-Driver 2026.03.19 shipped libnvidia-container 1.19.0 which has a regression in its prestart hook (ldcache error: process /sbin/ldconfig failed with error code: 1 → process confinement failed: invalid argument). Driver itself is fine, nvidia-smi works, only the legacy --gpus=all path is broken.

Workaround: switch the container to CDI.

1. On the host: nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml

2. In the Unraid template, change --gpus=all to --device=nvidia.com/gpu=all

3. Recreate the container

Driver stays on 595, no rollback risk.

https://www.reddit.com/r/unRAID/comments/1t0av6c/comment/oj9xm3j/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

1 hour ago, Beunhasa said:

I had the same issue with a 3060. I followed the following steps from a reddit comment which got the gpu to work:

That fixed it! Thank you so much. Was starting to pull my hair out troubleshooting this

its probably been asked and might not be plugin fault but can we stop the plugin from updating to latest version of driver despite hardware not being compatible with it? does it every unraid update, forcing follow up reboot hassle just to set driver back to what it should be.

as always, thanks for your work on the plugin.

Edited May 3May 3 by Mizerka

2 hours ago, Mizerka said:

its probably been asked and might not be plugin fault but can we stop the plugin from updating to latest version of driver despite hardware not being compatible with it? does it every unraid update, forcing follow up reboot hassle just to set driver back to what it should be.

as always, thanks for your work on the plugin.

Yes that's fixed for the next update.

Hi - after updating the plugin, then updating Unraid to 7.2.5, waiting for a reboot notification then rebooting, Plex isn't hardware transcoding (tested forcing 1080 file to 720 in browser). Was working before and haven't changed anything other than updating. Could have missed something in the last few pages here as it's got a bit confusing.

unraid1-diagnostics-20260503-1413.zip

7 minutes ago, Cessquill said:

Plex isn't hardware transcoding (tested forcing 1080 file to 720 in browser).

Please don't try it from the browser, this is a known issue with Nvidia and Plex that it will only show a black screen if you change the quality in the browser, please try a native app or set the quality directly in the settings to force a quality change when playback starts, this will also work in the browser, but please don't forget to reset that in the settings since you would always force a transcode if you leave it like that.

May I ask what script that is:

May 3 13:09:54 unRAID1 emhttpd: /usr/local/emhttp/plugins/user.scripts/backgroundScript.sh "/tmp/user.scripts/tmpScripts/Nvidia Powersave 2/script" >/dev/null 2>&1

I see also that you have this in your go file:

# nvidia-persistenced

From what I can see, the driver should work fine.

Can you please post your docker run command?

Hi - after updating the plugin, then updating Unraid to 7.2.5, waiting for a reboot notification then rebooting, Plex isn't hardware transcoding (tested forcing 1080 file to 720 in browser). Was working before and haven't changed anything other than updating. Could have missed something in the last few pages here as it's got a bit confusing.

unraid1-diagnostics-20260503-1413.zip

Weirdly my plex wasn't seeing or using the GPU despite getting my GPU working in Immich after some driver update shenanigans. To get Immich working I had to uninstall the driver, reboot, reinstall the driver, reboot, stop docker service, restart docker service. That was yesterday. Today I saw your post and chedked, boom Plex wasn't transcoding or even seeing the GPU. A restart of Plex just seemed to magically fix it. Crazy to think you haven't tried that already, but just in case you haven't [shrug]. I'm still on 7.2.4 also though.

4 minutes ago, jumperalex said:

Weirdly my plex wasn't seeing or using the GPU despite getting my GPU working in Immich after some driver update shenanigans. To get Immich working I had to uninstall the driver, reboot, reinstall the driver, reboot, stop docker service, restart docker service. That was yesterday. Today I saw your post and chedked, boom Plex wasn't transcoding or even seeing the GPU. A restart of Plex just seemed to magically fix it. Crazy to think you haven't tried that already, but just in case you haven't [shrug]. I'm still on 7.2.4 also though.

I haven't tried restarting Plex yet as a family member was using it (wasn't transcoding for them either). I will try shortly, but Plex can see the card for me. Sorry if I that sounds crazy to you.

Adding this also here. https://forums.unraid.net/topic/198521-unraid-os-version-725-available/page/2/#findComment-1621088
To sum up, the driver i need to have is v580.159.03, and it gets overwritten somehow with the latest every time i do an update, even though v580.159.03 is selected. Plugin is v 2026.05.01.

Also, the screen when you click update & download is blank, so i always have to download it manually via terminal with /usr/local/emhttp/plugins/nvidia-driver/include/download.sh

53 minutes ago, snoopy86 said:

To sum up, the driver i need to have is v580.159.03, and it gets overwritten somehow with the latest every time i do an update, even though v580.159.03 is selected. Plugin is v 2026.05.01.

The fix should work with the next Unraid upgrade, if not please report back (I've just pushed another fix in the Plugin Update Helper).

There might changed something but did you actually wait until it showed some text:

I just tried it and it works over here, granted that it took some time, however it looks like it's working as expected, do you possibly have an adblocker or something like that in use in your Browser :

My Nvidia drive updated tonight and now is not working.

The NVIDIA Quadro P2000 GPU installed in this system is

May 4 20:15:26 wserver1 kernel: NVRM: supported through the NVIDIA 580.xx Legacy drivers. Please

May 4 20:15:26 wserver1 kernel: NVRM: visit http://www.nvidia.com/object/unix.html for more

May 4 20:15:26 wserver1 kernel: NVRM: information. The 595.71.05 NVIDIA driver will ignore

May 4 20:15:26 wserver1 kernel: NVRM: this GPU. Continuing probe...

May 4 20:15:26 wserver1 kernel: NVRM: No NVIDIA GPU found.

Is there a path to downgrade to the last working driver?

Thanks for any help.

(I need to rebuild this ancient system, but it was running fine)

I figured out how to load an older drive, which I did and now I am back up.

Ross

On 5/2/2026 at 8:29 AM, Bluephoria said:

Despite updating to the latest plugin and swapping the driver, I'm still having issues.

Troubleshooting tried so far:

• Update plugin and reboot

• Upgrade driver (to 595) and reboot

• Uninstall driver, reboot, install driver, reboot

• Downgrade driver (to 580) and reboot
  

  I'm using an RTX 2000 Ada. The GPU ID is properly configured in my Plex and Tdarr dockers, but the GPU isn't able to be used.
  
  root@Storage:~# dmesg | grep -i nvidia

  [ 31.018588] nvidia: loading out-of-tree module taints kernel.

  [ 31.019314] nvidia: module license 'NVIDIA' taints kernel.

  [ 31.020648] nvidia: module license taints kernel.

  [ 31.231377] nvidia-nvlink: Nvlink Core is being initialized, major device number 239

  [ 31.240228] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem

  [ 31.289817] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.159.03 Fri Apr 24 06:16:47 UTC 2026

  [ 31.420257] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 580.159.03 Fri Apr 24 05:49:32 UTC 2026

  [ 31.423016] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver

  [ 31.424264] [drm] Initialized nvidia-drm 0.0.0 for 0000:02:00.0 on minor 1

  [ 83.268922] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.

The GPU is visible to the container:

root@Storage:~# docker exec -it Plex-Media-Server nvidia-smi

Sat May 2 06:27:18 2026

+-----------------------------------------------------------------------------------------+

| NVIDIA-SMI 580.159.03 Driver Version: 580.159.03 CUDA Version: 13.0 |

+-----------------------------------------+------------------------+----------------------+

| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |

| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |

| | | MIG M. |

|=========================================+========================+======================|

| 0 NVIDIA RTX 2000 Ada Gene... Off | 00000000:02:00.0 Off | Off |

| 30% 48C P0 17W / 70W | 0MiB / 16380MiB | 4% Default |

| | | N/A |

+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+

| Processes: |

| GPU GI CI PID Type Process name GPU Memory |

| ID ID Usage |

|=========================================================================================|

| No running processes found |

+-----------------------------------------------------------------------------------------+

Is there something obvious that I'm missing that would have been thrown out of whack with the latest unraid and driver updates? Anyone else having this experience?

I am also still having the same issue with same troubleshooting steps as the fixed build post and your troubleshooting steps too.

Do I need to change how the GPU is passed through in the template?

Edited May 5May 5 by trey81

UPDATE : FOUND THE ISSUE
Apparently, one of the VMs that uses the GPU autostarted with boot, and that caused the GPU to disappear completly because of the bypass .. I think the Nvidia plugin should always shows the card so we don't panic like that LOL.

///////////////
Alright, I think I might have a diffrent issue here I was on 7.2.4, both cards worked fine and I don't remember what the driver was

Then upgraded to 7.2.5 and both GPUs can be seen by the Nvidia driver, but got these errors

"NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

and this is in the logs "kernel: nvidia-nvlink: Unregistered Nvlink Core, major device number 240"

Both cards are shown in the NVIDIA driver page but they aren't working .. the driver page recommended downgrading from v595.71.05 to v580.159.03, which fixed my P400, but GTX 970 disappeared completely. I also can't go below v580.159.03.

Is this an issue with my Unraid, or shall I say goodbye to my GTX 970?

Because quick searching says that kernel updates with 7.2.5 caused this new behaviour, but I'm not sure about that.

warmind-diagnostics-20260505-1949.zip

NVIDIA Plugin is updated to the latest BTW

Edited May 5May 5 by TheExoticBeard
SOLVED

2 hours ago, TheExoticBeard said:

I think the Nvidia plugin should always shows the card so we don't panic like that LOL.

It can't, when passed through the card isn't visible to the host anymore. If it did it'd show you a card that's not there and it'd be even worse for troubleshooting...

18 hours ago, Ross said:

I figured out how to load an older drive, which I did and now I am back up.

Ross

I have the same NVIDIA Quadro P2000 GPU and seem to have the same issue.

Can anyone please give the steps to load an older driver that will work with this card?

EDIT: I am probably confused. After updating the Nvidia Driver plugin to 2026.05.01 running terminal, nvidia-smi reports driver version 575.51.02 so maybe that is ok.

For whatever reason it seems the card isn't being used for transcoding anymore by my jellyfin instance which was working before, but perhaps that is due to something else.

Edited May 5May 5 by dabl