[Plugin] Linuxserver.io - Unraid Nvidia


Recommended Posts

I have managed to get my GTX 1650 working once again. Unfortunately I don't know exactly what the fix was but on the off-chance someone else has issues where their card WAS working, but seemingly permanently broken after loading GUI mode (broken in the sense that the plugin couldn't detect it, at least), then here's what my final actions were before it started working again:

 

1. I powered down the server, with a clean shutdown.

2. I turned off the Power Supply, and waited for lights on the motherboard to fade. <-- This is the first time I did this. I tried restarting and shutting down the server many times during my investigating but only now decided to try removing all power for long enough for a "cold" start. I can't be sure but I do have my suspicions that this was the magic that fixed it. (Thanks to itimpi for this suggestion)

3. Relocated the graphics card to another slot. I had tried this before, but as mentioned above, only with the system off but the PSU still technically ON...

(Worth noting on this point that the graphics card had worked in either of the 2 available slots prior to me going into GUI mode, so I'm not convinced it's necessary)

4. Rebooted, obviously avoiding GUI mode like the black plague.

 

Thanks to those who offered suggestions

  • Like 1
Link to comment

When I used a quadro p2000, I was able to get the power output readings from nvidia-smi, which would pass that information along to wherever I wanted to view it - in this case Grafana.  When I tried to do the same with a quadro p1000, I get N/A for this field.  That seems odd, but it may be that it doesn't support this field?  I wasn't able to definitively find an answer for this. I was wondering if anyone could confirm this.  I talked to a friend that mentioned it may be the driver itself?  Thanks for any help!

 

I am using unraid nvidia 6.8.3 currently and the driver version is 440.59.

 

image.png.046f2c836293b80b2165b97d48eda4a7.png

Link to comment

 

I'd just like to add that i'm having the same issue as @Scroopy Noopers, but my card is running in a VM...so seems to be a wider issue. 

I have a EVGA GTX1070 FTW being passed from UnRAID 6.9.0 to Ubuntu 20.04

 

So far things i've tried: 

- Upgrading drivers from NVIDIA 450.66 to 455.38

- Upgrading the Ubuntu VM

- Upgrading UnRaid 6.9.0-beta29 to 6.9.0-beta30

- Checked Nouveau driver blacklisting

- Binding to VFIO-PCI

- Upgrading Machine version from Q35-4.2 to Q35-5.1

 

Quote

$ sudo lshw -C display
  *-display
       description: VGA compatible controller
       product: GP104 [GeForce GTX 1070]
       vendor: NVIDIA Corporation
       physical id: 0
       bus info: pci@0000:07:00.0

       version: a1
       width: 64 bits
       clock: 33MHz
       capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
       configuration: driver=nvidia latency=0
       resources: iomemory:80-7f iomemory:80-7f irq:22 memory:98000000-98ffffff memory:800000000-80fffffff memory:810000000-811ffffff ioport:7000(size=128) memory:99080000-990fffff
 

Quote

$ lspci | grep NVIDIA
07:00.0 VGA compatible controller: NVIDIA Corporation GP104 [GeForce GTX 1070] (rev a1)
07:00.1 Audio device: NVIDIA Corporation GP104 High Definition Audio Controller (rev a1)

Quote

$ dmesg | grep NVIDIA
[    3.547010] nvidia: module license 'NVIDIA' taints kernel.
[    3.709259] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  455.38  Thu Oct 22 06:06:59 UTC 2020
[    3.716272] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms  455.38  Thu Oct 22 05:57:59 UTC 2020

 

$ dmesg | grep NVRM
[    3.709259] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  455.38  Thu Oct 22 06:06:59 UTC 2020
[   12.496509] NVRM: GPU 0000:07:00.0: RmInitAdapter failed! (0x26:0xffff:1239)
[   12.496827] NVRM: GPU 0000:07:00.0: rm_init_adapter failed, device minor number 0

[   60.955912] NVRM: GPU 0000:07:00.0: RmInitAdapter failed! (0x26:0xffff:1239)
[   60.956055] NVRM: GPU 0000:07:00.0: rm_init_adapter failed, device minor number 0

Quote

$ sudo nvidia-settings -v

nvidia-settings:  version 455.38
  The NVIDIA X Server Settings tool.

Quote

$ sudo nvidia-smi
No devices were found

 

 

Link to comment

Hello!  I could use some help.  I am new to unRaid and am building a Plex Server and have a Quadro P2200 which I want to use for transcoding.  I installed the Unraid Nvidia Build 6.8.3 and it is not detecting my GPU. 

 

image.png.a4e3738a94406a49e7618c56a97c4c48.png

 

However, it does show up in the System Devices list:

image.png.3ac3fd86e927b6334174f25e4d62dec1.png

 

HVM and IOMMU is enabled

image.png.ee6cec8c9f3efab6c67ae121b576f412.png

 

As noted above, I have an Aorus Master x570 motherboard, Ryzen 3700x, 64GB Ram, LSI SAS Controller with 8 WD White Labels, (3) Samsung 970 NVMEs, and the Quadro P2200.  I have a VM installed, but it is not running and have only used it headless via VNC.

 

I tried to set it back to stock build and then reinstalled the Nvidia Build to no avail.  Even tried going up to newest beta.

 

Any help and guidance would be much appreciated.  So far, I am loving UnRaid - everything had been great except for this issue.

 

 

EDIT: I tossed in an old 1050ti and it recognized it right away.  So either an issue with the p2200 (although it works in windows) or it is a driver issue with the unraid nvidia build.

server-diagnostics-20201031-0213.zip

Edited by provoking
Link to comment
10 hours ago, Jarsky said:

 

I'd just like to add that i'm having the same issue as @Scroopy Noopers, but my card is running in a VM...so seems to be a wider issue. 

I have a EVGA GTX1070 FTW being passed from UnRAID 6.9.0 to Ubuntu 20.04

 

So far things i've tried: 

- Upgrading drivers from NVIDIA 450.66 to 455.38

- Upgrading the Ubuntu VM

- Upgrading UnRaid 6.9.0-beta29 to 6.9.0-beta30

- Checked Nouveau driver blacklisting

- Binding to VFIO-PCI

- Upgrading Machine version from Q35-4.2 to Q35-5.1

 

 

 

I'm not really how you are running your card, but if you just pass it through to a VM, you have no need for the nvidia drivers plugin, as you can't use both at the same time.

Link to comment
12 hours ago, saarg said:

I'm not really how you are running your card, but if you just pass it through to a VM, you have no need for the nvidia drivers plugin, as you can't use both at the same time.

I know but just pointing out there may be a larger Linux kernel issue thats affecting Scroopy, given that mine recently stopped working with the same issue. 

Mines been working in that setup for a year, and it just stopped working for no reason in the last week...only thing i'd changed up to that point was regular UnRAID and Ubuntu upgrades. The only thing that I can find that appears relevant is entries that say 'tainted kernel'

 

e.g

Quote

Oct 31 10:29:37 mediabox kernel: [    3.557346] nvidia: loading out-of-tree module taints kernel.
Oct 31 10:29:37 mediabox kernel: [    3.557352] nvidia: module license 'NVIDIA' taints kernel.
Oct 31 10:29:37 mediabox kernel: [    3.557352] Disabling lock debugging due to kernel taint
Oct 31 10:29:37 mediabox kernel: [    3.568325] nvidia: module verification failed: signature and/or required key missing - tainting kernel

 

Edited by Jarsky
Link to comment
19 hours ago, Jarsky said:

I know but just pointing out there may be a larger Linux kernel issue thats affecting Scroopy, given that mine recently stopped working with the same issue. 

Mines been working in that setup for a year, and it just stopped working for no reason in the last week...only thing i'd changed up to that point was regular UnRAID and Ubuntu upgrades. The only thing that I can find that appears relevant is entries that say 'tainted kernel'

 

e.g

 

Given that it is a tainted kernel, I'm thinking that my next step is to replace the unraid nvidia installation of beta 30 with the stock OS and see if the plugin registers the GPU (which I don't expect it to) and then re-install the unraid nvidia kernel and see if that works. 

Link to comment

Hi all,

 

I have built an Unraid server and so far, so good !

I just got a Quadro P2000 GPU to add to the current config, to support hardware transcoding in Plex.

I went through numerous threads, including this one of course, and I think I now have a pretty good idea of how to proceed thanks to all of you.

The Unraid Nvidia Plugin is installed.

Now that I'm ready to open the box, I just have a simple question. Should I :

install Nvidia Unraid build without the P2000 plugged-in, shutdown, install the P2000 physically, and then restart

- or shutdown, install the P2000, restart, install Nvidia Unraid build, and then reboot

- or the order doesn't matter ?

 

Thanks in advance for your help !

Edited by Gnomuz
typo
Link to comment
6 hours ago, Scroopy Noopers said:

Given that it is a tainted kernel, I'm thinking that my next step is to replace the unraid nvidia installation of beta 30 with the stock OS and see if the plugin registers the GPU (which I don't expect it to) and then re-install the unraid nvidia kernel and see if that works. 

I don't really see the point in doing that. A reboot would do the same as the OS is unpacked on each boot. So it's the same as reinstalling it.

And no, stock unraid will not make anything appear in the Nvidia plugin.

Link to comment
3 hours ago, Gnomuz said:

Hi all,

 

I have built an Unraid server and so far, so good !

I just got a Quadro P2000 GPU to add to the current config, to support hardware transcoding in Plex.

I went through numerous threads, including this one of course, and I think I now have a pretty good idea of how to proceed thanks to all of you.

The Unraid Nvidia Plugin is installed.

Now that I'm ready to open the box, I just have a simple question. Should I :

install Nvidia Unraid build without the P2000 plugged-in, shutdown, install the P2000 physically, and then restart

- or shutdown, install the P2000, restart, install Nvidia Unraid build, and then reboot

- or the order doesn't matter ?

 

Thanks in advance for your help !

It doesn't matter.

Link to comment
2 hours ago, galego100 said:

Hello ,

 

My Unraid is 6.7.2 , when i install the nvidia app, and i go the plugin it only show unraid version 6.8 and above, i see some tutorial on youtube and on them i saw 6.7.2 version... What i´m doing wrong ?

Thanks

I guess we delete older versions. So you should probably update to the recent version.

Link to comment

Yesh, I dont understand why they think that.

 

You only have to buy a license if your setup is currently using a pirated, hacked, or blacklisted license or subverting the license scheme in some illegal way. If you are, shame on you!

Edited by BRiT
Link to comment
On 11/2/2020 at 6:29 AM, Scroopy Noopers said:

Given that it is a tainted kernel, I'm thinking that my next step is to replace the unraid nvidia installation of beta 30 with the stock OS and see if the plugin registers the GPU (which I don't expect it to) and then re-install the unraid nvidia kernel and see if that works. 

I finally figured out my issue. 

You might want to check if you're booting UnRAID in UEFI or BIOS...you can check this in UnRAID GUI going to Main > Boot Device (Flash) > scroll down bottom to Server Boot Mode. 

If it say's UEFI, try booting to your BIOS and change the Boot Priority to USB: <Device Name> instead of UEFI: <Device Name>

After Reboot, check the Server Boot Mode now says "Legacy". 

See if the GPU works now. 

 

Turned out when I updated my server, I set my boot priority to UEFI by mistake. 

Edited by Jarsky
  • Like 1
Link to comment
8 hours ago, Jarsky said:

I finally figured out my issue. 

You might want to check if you're booting UnRAID in UEFI or BIOS...you can check this in UnRAID GUI going to Main > Boot Device (Flash) > scroll down bottom to Server Boot Mode. 

If it say's UEFI, try booting to your BIOS and change the Boot Priority to USB: <Device Name> instead of UEFI: <Device Name>

After Reboot, check the Server Boot Mode now says "Legacy". 

See if the GPU works now. 

 

Turned out when I updated my server, I set my boot priority to UEFI by mistake. 

My system says legacy. Unfortunately my Dell R510 doesn't seem to play nice with unraid when running in UEFI mode. I do appreciate you thinking of me once you found your solution though. 

 

This potential solution confuses me, as unraid still recognizes the GPU under the PCI device list, and is available for passthrough to VMs. So it doesn't seem to me that it would be an issue with the BIOS of my server so much as unraid nvidia itself. 

Link to comment
8 hours ago, Scroopy Noopers said:

My system says legacy. Unfortunately my Dell R510 doesn't seem to play nice with unraid when running in UEFI mode. I do appreciate you thinking of me once you found your solution though. 

 

This potential solution confuses me, as unraid still recognizes the GPU under the PCI device list, and is available for passthrough to VMs. So it doesn't seem to me that it would be an issue with the BIOS of my server so much as unraid nvidia itself. 

I believe its more to do with Secure Boot in UEFI in my instance. You perhaps could try booting a Linux Live distro to make sure your card is actually working as it should normally, that would rule out UnRAID incase potentially your card has gone faulty or have a power issue to the card. 

Link to comment
On 10/29/2020 at 4:13 PM, endiz said:

To answer my own question, looks like plex now requires nvidia drivers 450.66 and up for NVENC to work. Looks like i'll have to wait for 6.9.0 to go stable before updating my drivers.

While waiting for 6.9.0 to go stable, does it make sense to revert Plex back to a version before this requirement, or is that version to far back and would mess up the docker trying to revert back to it?  Sorry if this is a noob question, this all still feels new to me.

 

I had been wanting to installed a quadro for a couple months now, and I finally did today not realizing this was an issue.  I have my LSIO Plex docker on auto update so I passed that version by a couple of version, and was thinking about reverting it back before this requirement but wasn't sure if there was a reason not to.

Link to comment
  • trurl locked this topic
Guest
This topic is now closed to further replies.