Hello, after trying to update Unraid, I am getting this notification from the alert saying to post my screenshot here. Not sure what else to do. Diagnostics attached. Thanks

tower-diagnostics-20260712-1005.zip

On 6/25/2026 at 1:04 PM, Kilrah said:

One day you'll eventually have to either stop updating or replace the card, when that is nobody can know.

According to my logs, it's already happened for my 1060, which uses the same GPU as the Quadro P2000:

NVRM: The NVIDIA NVIDIA GeForce GTX 1060 6GB GPU installed in this system is

NVRM: supported through the NVIDIA 580.xx Legacy drivers. Please

NVRM: visit http://www.nvidia.com/object/unix.html for more

NVRM: information. The 610.43.03 NVIDIA driver will ignore

NVRM: this GPU. Continuing probe...

NVRM: No NVIDIA GPU found.

I'm gonna see if I can install a specific version or something..

On 6/25/2026 at 1:04 PM, Kilrah said:

One day you'll eventually have to either stop updating or replace the card, when that is nobody can know.

1 minute ago, Phil Merricks said:

I'm gonna see if I can install a specific version or something..

This was actually super easy. For the uninitiated like myself you can just choose a 580 driver version in the plugin settings page for the driver.

Hi @ich777 ,

Doing pre-purchase due diligence before buying an RTX 5060 Ti for local LLM (Ollama) + Frigate vision + Plex on my unRAID box, and wanted to check with you directly given you've mentioned owning Blackwell hardware yourself.

My setup: Currently on a GTX 1070 (Pascal), unRAID 7.2.5, Supermicro X11SSM-F / Xeon E3-1245 v5. Planning to replace it with a PNY RTX 5060 Ti 16GB (GB206).

What looked promising: SpaceInvaderOne's Feb 2026 "Paperless-ngx Part 3" tutorial shows a clean setup — RTX 5070 Ti (GB203), Open Source Driver v590.48.01, selected from a version-pinned list in the plugin, Update & Download, restart, done. No issues shown.

What gave me pause: I found a separate report of an RTX 5060 Ti (GB206 specifically) failing to initialize on unRAID 7.2.4, driver 595.71.05 open-source. lspci sees the card, kernel modules load (nvidia, nvidia_uvm, nvidia_drm, nvidia_modeset), but nvidia-smi reports "No devices were found." dmesg shows:

NVRM: GPU0 kbusVerifyBar2_GB202: MMUTest BAR2 readback VA = 0x0 returned garbage

NVRM: GPU0 RmInitNvDevice: *** Cannot initialize the device

NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x24:0x72:1344)

NVRM: GPU0 _kgspBootGspRm: unexpected WPR2 already up, cannot proceed with booting GSP

NVRM: GPU0 RmInitAdapter: Cannot initialize GSP firmware RM

I also found this same WPR2/GSP-boot error class as two separate open, unresolved issues on NVIDIA's own open-gpu-kernel-modules repo (on different hardware, not GB206) — and apparently this error family goes back to driver branches on Turing/Ampere too, so it looks like a known-but-intermittent NVIDIA firmware quirk rather than something specific to Blackwell, GB206, or your plugin.

Questions, if you have a minute:

1. Have you tested or heard reports of a GB206 (RTX 5060/5060 Ti) specifically on unRAID? Any known issues, or is it working cleanly for others?

2. Is there a specific driver version you'd currently recommend for a fresh GB206 install?

3. Any known BIOS-side mitigation for this GSP-init failure class — specifically Above 4G Decoding / Resizable BAR?

4. Is this the kind of thing your open-source driver CI work (I saw the PR building version-pinned nvos builds) would eventually help with, or is it purely upstream NVIDIA GSP firmware territory outside the plugin's control either way?

Haven't bought the card yet — trying to get real data before spending ~$600 rather than finding out the hard way. Appreciate anything you can share.

4 hours ago, sakh1979 said:

Haven't bought the card yet — trying to get real data before spending ~$600 rather than finding out the hard way. Appreciate anything you can share.

most informations you get are here when you read up.

RTX5080 user here and had no issues with open source drivers.

overall, i assume there are plenty RTX5060/Ti users around here also using NV drivers, if there would be a major issue you would find it here ;)

and yes, most likely if there are issues, its due hardware combos or bios settings and so on, sometimes trial & error.

on your intel platform it should be pretty straight forward, but nobody can tell ... and yes, in the end its the NV driver and not modified, just compiled for Unraids kernel.

so may go ahead and get your card and hopefully enjoy oob ;)

6 hours ago, sakh1979 said:

you've mentioned owning Blackwell hardware yourself

I didn't and I don't have a Blackwell card.

6 hours ago, sakh1979 said:

Have you tested or heard reports of a GB206 (RTX 5060/5060 Ti) specifically on unRAID? Any known issues, or is it working cleanly for others?

No.

6 hours ago, sakh1979 said:

Is there a specific driver version you'd currently recommend for a fresh GB206 install?

The Open Source one.

6 hours ago, sakh1979 said:

Any known BIOS-side mitigation for this GSP-init failure class — specifically Above 4G Decoding / Resizable BAR?

You've mentioned all things that you have to check, also make sure that you are on the latest BIOS version.

6 hours ago, sakh1979 said:

Is this the kind of thing your open-source driver CI work (I saw the PR building version-pinned nvos builds) would eventually help with, or is it purely upstream NVIDIA GSP firmware territory outside the plugin's control either way?

GSP is closed firmware and just added to the driver package, the same as it is with the libraries, they are all closed source. Only the driver module is open source.

ive got a 5060Ti 16gb on unraid 7.3.2, and i installed it last september. its worked flawlessly on whatever version unraid was back then. just install the nvidia open source driver and everything works.

On 2026/6/11 at PM8点17分, strenczjan said:

我的Nvidia T400显卡和音频也存在问题。当我关闭Windows 11虚拟机时，Unraid会卡死，因为它没有释放Nvidia音频进程。

I use my setup the same way you do; I have an RTX 5060 graphics card. Previously, after shutting down the Windows 11 VM, Unraid would successfully take over the 5060 card and manage its power state. However, recently, shutting down Windows 11 causes the host machine to freeze; the only way to shut it down properly is to set the sound card to "None" within the VM settings. I am certain everything worked fine before—I wonder if this issue was caused by an upgrade to the open-source Nvidia drivers.

Hi all.

I've just started with unraid and I'm having a nightmare with my 5070ti. I have installed the Nvidia driver plugin. I know the 610 driver is locked to 5000 series, so have setup the 580 driver.

But the GPU just will not work with it. Every time I run any load on the GPU, the you doesn't work and it seems to crash the web UI fully. All my containers still working though, so it's not the full os crashing.

I have spent hours on this trying to find a fix without any joy. I even took the GPU out and put into my main rig and tested heavily, it is 100% fine hardware wise. A 6 hour furmark test and gaming tests were all fine.

When the GPU driver is installed it does show the pcie as gen1 only, I have even set it to be fixed gen3 in bios but it's still showing as 1. I have a 5600g which is only capable of up to gen3 hence no gen4 or 5.

Anybody else has similar issues and found a remedy?

58 minutes ago, Acemodder said:

I know the 610 driver is locked to 5000 series

But that's what you say you have?

44 minutes ago, Kilrah said:

But that's what you say you have?

Sorry what do you mean? When I install the nvidia driver, it auto-installs the latest driver, but I swap it to download and update to 580 driver wihc should be correct for a blackwell card (5070ti). I have the drivers uninstalled currently, so my nvidia GPU isn't working properly (due to lack of drivers). But let me install it and show you some screenshots etc of the problems I'm getting.

So this is the process I did....

Installed the Nvidia driver package. It auto installs 610 driver.

I also disabled the docker as requested in the window.

I then downgraded/upgraded to driver version v595 which is the open source driver which is meant for Blackwell GPUs.

I rebooted my server to ensure the driver is loaded. As you can see it is, and the RTX5070ti is picked up by the O/S.

GPU stats plugin see's the GPU fine with no load.

At this point I enable Docker again in my unraid settings.

So I get after a short period of time, my WebUI dies, but docker containers still work in the background, plex/sabnzbd/sonarr/etc

But the only way to get my webUI back is to do a hard reset of the unraid PC. The GPU doesn't actually do anything in terms of processing, it's just showing no usage etc, it's like the OS refuses to the use the GPU.

Restting the WebUI services with this works via ssh command line, so the server is actually working, just the GPU crash locks the webui service.

/etc/rc.d/rc.php-fpm restart

/etc/rc.d/rc.nginx restart

Oh... I finally think I fixed it.

I changed my BIOS primary GPU to be ipgu and not the nvidia RTX5070ti, seems that has resolves it for now....

I will continue to monitor it though.

NVIDIA CDI/Legacy hooks fail with "cannot open /proc/self/fd/N" - Apears to be positional bug runc, not driver-specific.

Following up on my earlier posts in this thread (dual RTX 3060 setup) — I did a deep isolation test and found something that should help pinpoint the root cause.

Environment

• Unraid 7.2.5 (Docker 29.5.3, runc 1.3.5 — commit v1.3.5-0-g488fc13)

• Nvidia-Driver plugin, tested with three different driver versions: 610.57.04, then downgraded to 580.173.02 — same result on all

• 2x NVIDIA GeForce RTX 3060 (GA106), 01:00.0 and 03:00.0

• Host: nvidia-smi works perfectly, both GPUs detected and idle

Error:

Every method of GPU injection into a container fails with the exact same error:

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 running createContainer hook #3: exit status 2, stdout: , stderr: /bin/sh: 0: cannot open /proc/self/fd/12: No such file exit status 2

Tested and all failed identically:

• --gpus all

• --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=all -e NVIDIA_DRIVER_CAPABILITIES=all

• --device=nvidia.com/gpu=all (CDI)

• --device=nvidia.com/gpu=all --security-opt seccomp=unconfined

• no-pivot = true added under [nvidia-container-cli] in /etc/nvidia-container-runtime/config.toml

None of the above changed the outcome. Downgrading the driver three times did not help either.

Isolation test

I inspected the generated /etc/cdi/nvidia.yaml and counted the createContainer hooks in the order they're combined for the CDI device requested:

• --device=nvidia.com/gpu=all → hook #3 (0-indexed) lands on the big create-symlinks hook (the one linking ~30 .so files: libcuda.so, libnvidia-ml.so, etc.)

• --device=nvidia.com/gpu=0 (single GPU only, fewer total hooks) → hook #3 lands on a completely different hook: enable-cuda-compat

Both failed with the identical /proc/self/fd/12 error, at the same hook index, despite the hook content being different. it's runc failing deterministically once you reach the 4th createContainer hook execution in a sequence, regardless of content.

Workaround that worked

Manually edited /etc/cdi/nvidia.yaml to strip it down to just one hook (the essential create-symlinks for the core .so files), removing:

• per-device create-symlinks (/dev/dri/by-path links)

• the small gbm/glxserver symlink hook

• enable-cuda-compat

• update-ldcache

• disable-device-node-modification

With only 1 hook left, docker run --rm --device=nvidia.com/gpu=all ubuntu nvidia-smi succeeded, both GPUs showing correctly.

However, this breaks anything relying on ldconfig cache inside the container — confirmed when testing h264_nvenc via jrottenberg/ffmpeg:4.1-nvidia, which failed with Cannot load libcuda.so.1 (dynamic loader couldn't find the symlinked lib without an updated ldcache). Adding -e LD_LIBRARY_PATH=/usr/lib64 as a workaround for that specific case.

Given that Unraid 7.2.5 shipped a runc update specifically patching CVE-2025-31133, CVE-2025-52565, and CVE-2025-52881 (procfs write-redirection / hook-related fixes), and the error signature (cannot open /proc/self/fd/N) matches exactly the kind of fd-handling path touched by that patch, I suspect there's a regression introduced by the runc hardening that breaks legitimate sequential OCI hook execution somewhere around the 4th hook.

Question:
Has anyone else hit this exact createContainer hook #N: /proc/self/fd error after updating to Unraid 7.2.5+ / runc 1.3.5?
Is there a known-good runc version pre-CVE-fix vs. post-fix that avoids this regression, or is this a brand new bug specific to runc 1.3.5?

has the nvidia driver broke... it was working last week.. yesterday my jellfin stopped working after a reboot found nvidia driver cant find my video card..

i tried to re download update.. it says its downloading but actually its not downloading anything... after a long time it says cant download... i uninstalled the plugin rebooted.. reinstalled the plugin i check my pfsense and the unraid server is not downloading the nvidia driver... is there something broken in the past week.

4 hours ago, comet424 said:

is there something broken in the past week.

by the mass using this i would rather say no ;) there woul dbe more posts about this.

most likely, dns (provider) issue for git access or broken USB devices, or not enough space (rather rare), and so on.

may test if your docker updates are still working, may try a VPN to bypass ...

If anyone is still struggling with this after updating from <= 7.2.4 to anything above 7.2.5 here is what worked for me.

To begin with this was the error when trying to start the Emby Docker Container:
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: mount dst=/proc, flags=MS_REC|MS_SLAVE: invalid argument

The Share Mounts defined in the container were all defined as Read/Write (Shared).
All I had to do is change all of them to Read/Write (Slave)

Without re posting everything in its raw format, would anyone be able to shed some light on the following issue I'm having adding a 2nd GPU to my system and if the family of GPU would be the likely cause of my 2nd GPU not showing up?

Adding 2nd GPU not showing to Unraid - General Support - Unraid

TLDR - Have a P4, got a T4 but both dont seem to work together only 1 appears in the plugin while the other wont show but is detected by Unraid

6 hours ago, brent3000 said:

TLDR - Have a P4, got a T4 but both dont seem to work together only 1 appears in the plugin while the other wont show but is detected by Unraid

Your GPU fell from the BUS, could be caused by anything:

Sep  4 15:21:17 DellBox kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 237
Sep  4 15:21:17 DellBox kernel: 
Sep  4 15:21:17 DellBox kernel: nvidia 0000:04:00.0: Unable to change power state from D3cold to D0, device inaccessible
Sep  4 15:21:17 DellBox kernel: NVRM: The NVIDIA GPU 0000:04:00.0
Sep  4 15:21:17 DellBox kernel: NVRM: (PCI ID: 10de:1eb8) installed in this system has
Sep  4 15:21:17 DellBox kernel: NVRM: fallen off the bus and is not responding to commands.

Please make sure that you update your BIOS, enable Resizable BAR support and Above 4G decoding. Make also sure that you've booting into UEFI Mode.

You could also try another PCIe slot if you have on available but I'm not super sure if that's a Dell artificial limitation, you'll never know with Dell hardware.

Hello, I'm am having issues with this after trying to upgrade unraid to 7.3.2. Here are the details:

Nvidia Driver plugin selects wrong version on Unraid 7.3.2 upgrade

System: Unraid 7.3.2, NVIDIA T400 4GB (TU117GL / Turing), driver plugin currently on 580.126.09 (not yet rebooted into 7.3.2).

Steps:

1. Started the 7.3.2 upgrade (not yet rebooted). Got the "unRAID version change detected" notification, followed by:

   • Nvidia Driver v580.173.02 download failed

   • Download from plugin package(s): Nvidia Driver for unRAID v7.3.2 failed!

2. Checked the Nvidia Driver plugin page. Plugins page shows "up to date," but the driver info section lists:

   • Supported candidates: 580.159.03, 590.48.01, 595.71.05

   • Recommended: 595.71.05 (open-source)

3. Selected the 595.71.05 radio button and triggered a download. Log output shows it attempting to download v575.64.05 instead — a version that isn't even in the supported candidates list.

4. Hard-refreshed (and tried different browser) and re-selected 595.71.05 — same result, still tries 575.64.05.

Question: Is this a known issue with the version selector not saving/applying on this build? Is there a workaround (e.g. manually editing settings.cfg) to force the correct version, or should I wait for a plugin update before rebooting into the new kernel?

Diagnostics attached if needed. Haven't rebooted yet — holding off until this is sorted since 575.64.05 isn't confirmed built for the new kernel.

ds9-diagnostics-20260907-1054.zip