[Plugin] Nvidia-Driver

[ich777]

4 hours ago, UnJustice said:

Well, I'll look for your reply and again I appreciate your time and help.

I created an Issue on GitHub over here, I hope that's okay for you and that I included a few files from the Diagnostics.

 

4 hours ago, UnJustice said:

Also, I used a totally different Plex docker image.

I always recommend to use the official image for testing purposes if you experience such issues.

[ich777]

On 8/23/2023 at 4:14 AM, UnJustice said:

Short version: The Krusader docker is causing it for some reason. I understand this isn't the place to ask for support on that, so I'm not. I'm just simply reporting that it was the thing causing the runc error.

Can you please provide a screenshot from the Docker template from this container?

 

On 8/23/2023 at 4:14 AM, UnJustice said:

Krusader was added and immediately when I stopped Plex and tried to restart it blam-o! I got the error again.

Which Krusader container are you using (repository)? Have you yet tried another Krusader container from another maintainer?

 

BTW Thanks for the detailed explanation, this might be really useful in the future.

[warzonefury]

Hello everyone!
 

Since yesterday, I've been trying to get hardware transcoding to work with my RTX but I can't. I tried with the official version of plex and also with the linuxserver version.
 

In both cases, when I launch a video and I lower the quality so that it transcodes, the video continues until the end of the buffering and loads continuously. It doesn't transcode.
 

I can't see where I'm wrong. I followed the guide of Ich777 and other indications found on the web but whatever, it does not work.
 

Here are some images that may help you find the solution, thank you

[ich777]

37 minutes ago, warzonefury said:

I can't see where I'm wrong. I followed the guide of Ich777 and other indications found on the web but whatever, it does not work.

I would recommend that you first try it with the official Plex container.

 

How are you trying to transcode? Please note that the web client from Plex has issues when enabling transcoding.

 

The driver is working properly from what I can see. The logs from Plex, especially the transcoding log would be helpful but that is something for the Docker container thread.

 

Again, I would first recommend that you try the official Plex container.

[warzonefury]

13 minutes ago, ich777 said:

How are you trying to transcode? Please note that the web client from Plex has issues when enabling transcoding.

That was it ! I was trying to transcode by the browser, I just tried with the application, it works flawlessly. But, the browser bug is recent? Is it related to Unraid? It's still quite annoying

[ich777]

13 minutes ago, warzonefury said:

Is it related to Unraid?

Why should it be related to Unraid?

 

You are using a Docker container which is isolated from the Host, you are using a browser which is running not on Unraid, you are using the Plex web client.

Hope that explains it a bit better. 😉

 

Please report that on the Plex forums if you want a fix for that.

[warzonefury]

3 hours ago, ich777 said:

Why should it be related to Unraid?

 

You are using a Docker container which is isolated from the Host, you are using a browser which is running not on Unraid, you are using the Plex web client.

Hope that explains it a bit better. 😉

I expressed myself badly, I mean, it comes from the plex container on unraid in general or just isolated for me? On the web app, does transcoding work for you?

[ich777]

12 minutes ago, warzonefury said:

it comes from the plex container on unraid

This is an issue from Plex on Linux, in general and AFAIK knowm by the developers, I think if you read back one or two pages you‘ll find a link to the Plex forums where multiple users reported that.

 

So to speak this affects all Linux users.

 

12 minutes ago, warzonefury said:

On the web app, does transcoding work for you?

Yes, but only because I set a limit for the maximum allowed bitrate because I don‘t use Plex, only for testing if the driver works.

(If you set a general limit it is working but not if you are changing it on the fly)

 

I‘m a Emby guy.

[warzonefury]

44 minutes ago, ich777 said:

This is an issue from Plex on Linux, in general and AFAIK knowm by the developers, I think if you read back one or two pages you‘ll find a link to the Plex forums where multiple users reported that.

 

So to speak this affects all Linux users.

 

Yes, but only because I set a limit for the maximum allowed bitrate because I don‘t use Plex, only for testing if the driver works.

(If you set a general limit it is working but not if you are changing it on the fly)

 

I‘m a Emby guy.

 

Thanks for all informations ! ❤️

[Sideh]

Hello, I am hoping to get some help or guidance with an issue. I had my GPU running in a previous system with unraid but recently bought an old T630 and installed my GPU (GTX 1070) into it with the power adapter provided from the seller. It lights up and seems to be working but after installing the app i get an error.

 

Quote

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

 

Is there anything I can do? I have done the following:

 

• made sure the PCIe slot is enabled in the bios settings.
• tried older driver versions.
• uninstalled and reinstalled the app.

Now I am scratching my head wondering what other options I have.

 

Sincerely,

Sideh.

 

PS. doing the nvidia-smi in console provides the same out put as in the picture.

[ich777]

8 minutes ago, Sideh said:

Is there anything I can do?

Yes, post your Diagnostics here.

 

 

9 minutes ago, Sideh said:

an old T630 and installed my GPU (GTX 1070) into it with the power adapter provided from the seller

I'm not sure what you are talking about...

[Sideh]

12 hours ago, ich777 said:

Yes, post your Diagnostics here.

I am so sorry, can't believe I missed that. Here it is attached.

 

12 hours ago, ich777 said:

I'm not sure what you are talking about...

Just mentioning I am using and old Dell server incase that was an issue. Just incase.

daedalus-gaming-diagnostics-20230829-1511.zip

[ich777]

2 hours ago, Sideh said:

I am so sorry, can't believe I missed that. Here it is attached.

Your card is bound to VFIO:

84:00.0 VGA compatible controller [0300]: NVIDIA Corporation GP104 [GeForce GTX 1070] [10de:1b81] (rev a1)
    Subsystem: ASUSTeK Computer Inc. GP104 [GeForce GTX 1070] [1043:8599]
    Kernel driver in use: vfio-pci
    Kernel modules: nvidia_drm, nvidia
84:00.1 Audio device [0403]: NVIDIA Corporation GP104 High Definition Audio Controller [10de:10f0] (rev a1)
    Subsystem: ASUSTeK Computer Inc. GP104 High Definition Audio Controller [1043:8599]
    Kernel driver in use: vfio-pci

 

 

Please unbind it from VFIO and try again, this is the entry:

Loading config from /boot/config/vfio-pci.cfg
BIND=0000:84:00.0|10de:1b81 0000:84:00.1|10de:10f0---
Processing 0000:84:00.0 10de:1b81
Vendor:Device 10de:1b81 found at 0000:84:00.0

IOMMU group members (sans bridges):
/sys/bus/pci/devices/0000:84:00.0/iommu_group/devices/0000:84:00.0
/sys/bus/pci/devices/0000:84:00.0/iommu_group/devices/0000:84:00.1

 

[Sideh]

omg I can't believe I missed that. I untick it multiple times and then scrolled to the bottom and hit done. My bad, Sorry about that.

 

It now says "No devices were found" I have also verified that it is no longer selected for VFIO.

 

I have also included the diagnostics this time 😅

daedalus-gaming-diagnostics-20230829-1756.zip

[ich777]

31 minutes ago, Sideh said:

omg

Can you spot the issue:

Aug 29 00:54:40 Daedalus-Gaming kernel: NVRM: GPU 0000:84:00.0: GPU does not have the necessary power cables connected.
Aug 29 00:54:40 Daedalus-Gaming kernel: NVRM: GPU 0000:84:00.0: RmInitAdapter failed! (0x24:0x1c:1436)
Aug 29 00:54:40 Daedalus-Gaming kernel: NVRM: GPU 0000:84:00.0: rm_init_adapter failed, device minor number 0

 

[Sideh]

4 minutes ago, ich777 said:

Can you spot the issue:

Aug 29 00:54:40 Daedalus-Gaming kernel: NVRM: GPU 0000:84:00.0: GPU does not have the necessary power cables connected.
Aug 29 00:54:40 Daedalus-Gaming kernel: NVRM: GPU 0000:84:00.0: RmInitAdapter failed! (0x24:0x1c:1436)
Aug 29 00:54:40 Daedalus-Gaming kernel: NVRM: GPU 0000:84:00.0: rm_init_adapter failed, device minor number 0

 

Well not enough power. Thanks for clearing that up. I have the power cables connected but it musn't be enough. Thanks for clearing that up for me. Apologies for wasting your time on this.

 

EDIT: Spelling

Edited August 29, 2023 by Sideh

[ich777]

4 minutes ago, Sideh said:

power cables

Please check if the are plugged all the way in.

[KrypticKahos]

I previously had a 1050Ti GPU installed without issues being used with dockers for transconding. I just put in a 3050 and get no device found.

 

nvidia-smi output gives "No Devices were Found"

VFIO is unbound.

Graphics card seems functional as I have video out on the card during bootup.

Diagnostics are attached.

 

I have also tried multiple driver versions, allowing it to fully download and install and rebooted.

 

Have any other input on why it may not be visible?

sv1-diagnostics-20230829-1543.zip

[ich777]

9 hours ago, KrypticKahos said:

Have any other input on why it may not be visible?

Please update your BIOS, Asrock had some issues with GPU compatibility I've read in a article but that was for RTX40 series but maybe it helps in your situation too.

 

I can only see this messages over and over:

Aug 29 15:30:54 SV1 kernel: NVRM: GPU 0000:04:00.0: RmInitAdapter failed! (0x25:0x65:1462)
Aug 29 15:30:54 SV1 kernel: NVRM: GPU 0000:04:00.0: rm_init_adapter failed, device minor number 0

 

This is a pretty generic message and often times means that there is a hardware incompatibility issue.

For earlier Ryzen CPUs I always recommended disabling C-States in the BIOS or forcing a specific PCIe generation in the BIOS instead of AUTO but please try upgrading the BIOS first and see if that helps.

[KrypticKahos]

15 hours ago, ich777 said:

Please update your BIOS, Asrock had some issues with GPU compatibility I've read in a article but that was for RTX40 series but maybe it helps in your situation too.

 

I can only see this messages over and over:

Aug 29 15:30:54 SV1 kernel: NVRM: GPU 0000:04:00.0: RmInitAdapter failed! (0x25:0x65:1462)
Aug 29 15:30:54 SV1 kernel: NVRM: GPU 0000:04:00.0: rm_init_adapter failed, device minor number 0

 

This is a pretty generic message and often times means that there is a hardware incompatibility issue.

For earlier Ryzen CPUs I always recommended disabling C-States in the BIOS or forcing a specific PCIe generation in the BIOS instead of AUTO but please try upgrading the BIOS first and see if that helps.

BIOS was updated, and you were correct. The notes for the new version had fixes for 40 series card compatibility. This however didn't resolve my issues.

 

I will also go and attempt disabling C-states, and forcing the PCIe generation.

[KrypticKahos]

4 hours ago, KrypticKahos said:

BIOS was updated, and you were correct. The notes for the new version had fixes for 40 series card compatibility. This however didn't resolve my issues.

 

I will also go and attempt disabling C-states, and forcing the PCIe generation.

 

Did some more testing and C-states were already disabled, and setting PCIe generation didn't help either.

 

Turns out it was because I had an odd hardware configuration. I had my GPU in the secondary PCIe slot, and my HBA card in the primary GPU slot. I did this previously because I wanted maximum bandwidth to the hard drives. Turns out that my motherboard didn't like that setup. I put the GPU in the primary slot, and HBA in the secondary and now the GPU is showing up and everything is working as expected.

[capino]

Just updated the plugin to the latest version (2023.08.31)

 

But cannot update to any driver version anymore.
There is a deprivation notification now.

 

 

When I try to update to latest I get the following message:

[ich777]

@ezhik & @giafidis thanks for the report.

I've now update the plugin_update_helper which is part of most third party driver plugins.

The upgrade process should now work as intended when you have set the Production Branch or New Feature Branch in the plugin and should not switch back to the Latest branch (which is currently also the Production Branch version).

 

The background to why it fails is that version 535.98 doesn't follow the usual versioning scheme from Nvidia and is missing .xx and comm can't compare if the versions numbers are not the same scheme.

Made a update to the plugin a few months ago but completely forgot to also update the plugin_update_helper, sorry...

 

BTW Please always report such issues in the appropriate support thread from the plugin, because the Nvidia Driver is strictly speaking not part of Unraid since it's a third party plugin.

@limetech included graciously a check for the plugin_update_helper to make sure that users are on the latest version, even if it's not part of Unraid.

 

 

Anyways, here are the tests with the updated version:

Downgrade to 6.12.13 and set the version to Production Branch (please ignore that no card is found because I currently have no card in my Test server) :

 

 

Pull the upgrade over the built in updater:

 

 

Wait for the notifications to say it is safe to reboot:

 

 

After the reboot check if it still at Production Branch:

[ich777]

3 minutes ago, capino said:

But cannot update to any driver version anymore.

Please try if you can download that file from your local PC: Click

 

Nothing has changed in terms of the download routine.

 

Please uninstall the plugin, reinstall it and see if that helps (a reboot in between would be also recommended).

[capino]

1 hour ago, ich777 said:

Please try if you can download that file from your local PC: Click

 

Nothing has changed in terms of the download routine.

 

Please uninstall the plugin, reinstall it and see if that helps (a reboot in between would be also recommended).

Reinstalling the plugin seems to resolve the problem. (without reboot in between)

During the reinstall, driver version v535.104.05 has been downloaded.
I still have to reboot the server. This will be done somewhere this weekend since I'm not at home at the moment.

Edited September 1, 2023 by capino