[Plugin] Nvidia-Driver

[ich777]

7 minutes ago, Little709 said:

Although the error does persist when using the windows client. Which is obviously quite weird. But it seems that it's indeed not related to either the plugin or the container. Thank you for the help though..

Do you only tested the Plex web client or any native client too, completely forgot that sometimes the web client doesn't initialize the HW transcoding properly.

[Ichithekiller]

Hey all. Having an issue with Nvidia drivers. Running 6.9.1
Recently got VMs working with a Quadro m2000 card. Was working great. A new update for the drivers popped up, so I tried to install it. Wouldn't install and says this when I try to install:
-----ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR------
----Can't get Production Branch version and found no installed local driver-----
-----Please wait for an hour and try it again, if it then also fails please-----
------go to the Support Thread on the Unraid forums and make a post there!------
Tried again later a few times. Lo luck. VMs are not working now.
Tried to uninstall the plugin to reinstall to see if that fixed it. it uninstalled.. kinda. Still shows up under the settings page, but isn't listed on the plugins page. Under the settings page I click Nvidia driver and it doesn't show my card anymore.
Trying to reinstall says this:
First this pops up with an ok/cancel:
This application has additional requirements
Compatible nVidia GPU
You hit ok, then it says this this:
plugin: installing: https://github.com/.../unrai.../raw/master/nvidia-driver.plg
plugin: downloading https://github.com/.../unrai.../raw/master/nvidia-driver.plg
plugin: downloading: https://github.com/.../unrai.../raw/master/nvidia-driver.plg ... done

+==============================================================================
| Skipping package nvidia-driver-2022.01.20 (already installed)
+==============================================================================

-----ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR------
----Can't get Production Branch version and found no installed local driver-----
-----Please wait for an hour and try it again, if it then also fails please-----
------go to the Support Thread on the unRAID forums and make a post there!------
plugin: run failed: /bin/bash retval: 1

Updating Support Links

Finished Installing. If the DONE button did not appear, then you will need to click the red X in the top right corner

 

Not sure what to do at this point.. is there a way to manually remove the plugin in unRAID?
 

[ich777]

9 minutes ago, Ichithekiller said:

VMs are not working now.

What do you mean with that?

The plugin isn't needed for VMs... (please see the first post in this thread)

 

9 minutes ago, Ichithekiller said:

Can't get Production Branch version and found no installed local driver

Are you able to connect to Github from anywhere in you local network? Does your server have a active internet connection? Do you have any AdBlocking software installed somewhere in your network, if yes try to disbale it at least 10 minutes before you try to download the plugin.

 

9 minutes ago, Ichithekiller said:

Not sure what to do at this point.. is there a way to manually remove the plugin in unRAID?

Have you tried to reboot after it failed to install or tried to reboot after uninstalling it?

[Ichithekiller]

On 1/30/2022 at 11:24 PM, ich777 said:

What do you mean with that?

The plugin isn't needed for VMs... (please see the first post in this thread)

Sorry, should have specified. The vm starts, but there is no video output from my video card now. When I go to Nvidia drivers in the settings screen it doesn't have my card listed anymore, but I can select the card from the edit vm page. 

On 1/30/2022 at 11:24 PM, ich777 said:

 

Are you able to connect to Github from anywhere in you local network? Does your server have a active internet connection? Do you have any AdBlocking software installed somewhere in your network, if yes try to disbale it at least 10 minutes before you try to download the plugin.

Server has access to internet. Able to update other plugins/ docker containers with no problem. No add blocking software. 

On 1/30/2022 at 11:24 PM, ich777 said:

 

Have you tried to reboot after it failed to install or tried to reboot after uninstalling it?

Have rebooted it with no change for both instances.

 

Is there a way to manually delete a plugin so I can try to reinstall again?

[itimpi]

11 minutes ago, Ichithekiller said:

Is there a way to manually delete a plugin so I can try to reinstall again?

Each plugin will have a .plg file in the config folder on the flash drive and (typically) an associated folder for holding plugin related files.   Deleting these and rebooting means the plugin will no longer be present.

[localh0rst]

Hi All,

I'm in the situation where I tried the plugin - worked very well (thanks for that!) but I wanted to go back and passthrough the GPU (RTX3070) in a VM, unfortunately, I don't see any PCIe devices available for pass through - I am very sure the GPU showed up before. Is this something the driver might have caused or do I have to look somewhere else? Tried to mess around with 

append vfio-pci.ids=10de:2484,10de:228b

into syslinux.cfg, but no luck until now. Any help would be appreciated! Thanks!

localserver-diagnostics-20220201-1925.zip

Edited February 1, 2022 by localh0rst
added logs

[ich777]

1 hour ago, Ichithekiller said:

Is there a way to manually delete a plugin so I can try to reinstall again?

Do this from a unRAID terminal:

rm -R /boot/config/plugins/nvidia-driver /boot/config/plugins/nvidia-driver.plg

 

1 hour ago, Ichithekiller said:

Server has access to internet. Able to update other plugins/ docker containers with no problem. No add blocking software. 

What happens when you press "Download" on the plugin page or do you have no plugin page at all in the settings or can't click the icon?

 

Please share your Diagnostics.

[ich777]

40 minutes ago, localh0rst said:

I wanted to go back and passthrough the GPU (RTX3070) in a VM, unfortunately, I don't see any PCIe devices available for pass through

Uninstall the plugin, remove "vfio-pci.ids=10de:2484,10de:228b" from your syslinux.config, go to Tools -> System devices, check the boxes at your Nvidia card (don't forget to also check the box for the Audio device from the card and click on Bind to VFIO and reboot after that.

If you can't bind it to VFIO reboot and try to do the steps from above.

 

That is more a VM related question...

[Ichithekiller]

4 hours ago, ich777 said:

Do this from a unRAID terminal:

rm -R /boot/config/plugins/nvidia-driver /boot/config/plugins/nvidia-driver.plg

Just tried that. It says "no such file or directory"

4 hours ago, ich777 said:

What happens when you press "Download" on the plugin page or do you have no plugin page at all in the settings or can't click the icon?

 

Please share your Diagnostics.

Nothing under the plugin page. under settings it has the Nvidia driver page. 

I go to the settings page and click Nvidia drivers. Hit download and it says this:

"-----ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR - ERROR------
----Can't get Production Branch version and found no installed local driver-----
-----Please wait for an hour and try it again, if it then also fails please-----
------go to the Support Thread on the Unraid forums and make a post there!------"

 

Here is the diagnostics report. 

ichithetower-diagnostics-20220201-1859.zip

[Ichithekiller]

8 hours ago, ich777 said:

Do this from a unRAID terminal:

rm -R /boot/config/plugins/nvidia-driver /boot/config/plugins/nvidia-driver.plg

 

What happens when you press "Download" on the plugin page or do you have no plugin page at all in the settings or can't click the icon?

 

Please share your Diagnostics.

 

Ended up doing what itimpi said and deleted the plugin and matching folder from the flash drive. Was able to reinstall the plugin. Plugin shows up under both settings and plugin screens again. My Quadro m2000 shows up under installed GPUs like before. Still no video output on already made (and previously working) VMs unfortunately though. Not sure why..

[alturismo]

1 hour ago, Ichithekiller said:

My Quadro m2000 shows up under installed GPUs like before. Still no video output on already made (and previously working) VMs unfortunately though. Not sure why..

 

may to clarify, you are talking about your GPU usage in a VM, therefore you

 

- dont need this plugin, the nvidia-driver plugin is intended for Nvidia docker usage on unraid, not for VM's

- you should read up, use the VFIO bind in tools, system devices, assign it to your VM and thats it

 

 

then the card should be free and so on ... thats why its not really the right place here as this plugin is for a different usecase

[ich777]

5 hours ago, Ichithekiller said:

Just tried that. It says "no such file or directory"

2 hours ago, Ichithekiller said:

Ended up doing what itimpi said and deleted the plugin and matching folder from the flash drive.

The plugin was uninstalled from what I see and then you tried it to install it again and also you've hit the Download button from the plugin page a few times from what I see in your log.

 

2 hours ago, Ichithekiller said:

Plugin shows up under both settings and plugin screens again. My Quadro m2000 shows up under installed GPUs like before.

But the Diagnostics are from before it shows up in the plugin and the plugin is successfully installed or am I wrong, the Diagnostics show that the plugin isn't installed but your card shows up in your System Devices just fine:

02:00.0 VGA compatible controller [0300]: NVIDIA Corporation GM206GL [Quadro M2000] [10de:1430] (rev a1)
	Subsystem: Hewlett-Packard Company GM206GL [Quadro M2000] [103c:1190]
02:00.1 Audio device [0403]: NVIDIA Corporation GM206 High Definition Audio Controller [10de:0fba] (rev a1)
	Subsystem: Hewlett-Packard Company Device [103c:1190]

 

2 hours ago, Ichithekiller said:

Still no video output on already made (and previously working) VMs unfortunately though. Not sure why..

I think you are talking about showing up on the VM template page or am I wrong? If you only want to use this card in a VM then follow the instructions from @alturismo and you should be good to go, also see this note about the plugin on the first page:

 

 

But you want to use the plugin in a container too I think, at least the Diagnostics show this but also keep in mind that I don't recommend using it this way because you can have other issues like the one you described here to.

[Ichithekiller]

9 hours ago, ich777 said:

The plugin was uninstalled from what I see and then you tried it to install it again and also you've hit the Download button from the plugin page a few times from what I see in your log.

 

But the Diagnostics are from before it shows up in the plugin and the plugin is successfully installed or am I wrong, the Diagnostics show that the plugin isn't installed but your card shows up in your System Devices just fine:

02:00.0 VGA compatible controller [0300]: NVIDIA Corporation GM206GL [Quadro M2000] [10de:1430] (rev a1)
	Subsystem: Hewlett-Packard Company GM206GL [Quadro M2000] [103c:1190]
02:00.1 Audio device [0403]: NVIDIA Corporation GM206 High Definition Audio Controller [10de:0fba] (rev a1)
	Subsystem: Hewlett-Packard Company Device [103c:1190]

 

I think you are talking about showing up on the VM template page or am I wrong? If you only want to use this card in a VM then follow the instructions from @alturismo and you should be good to go, also see this note about the plugin on the first page:

 

 

But you want to use the plugin in a container too I think, at least the Diagnostics show this but also keep in mind that I don't recommend using it this way because you can have other issues like the one you described here to.

Diagnostics report was before I manually deleted it from the flash drive. After I did that, it installed with no issue. 

 

Ah. Got ya. I did not realize the plugin was only for dockers. Didn't read the fine print, just assumed I needed it for a card to work. Sorry for the confusion. I'm still fairly new to unraid and VERY new to VMs. I will give VFIO bind a try. Thanks!

[Ichithekiller]

11 hours ago, alturismo said:

 

may to clarify, you are talking about your GPU usage in a VM, therefore you

 

- dont need this plugin, the nvidia-driver plugin is intended for Nvidia docker usage on unraid, not for VM's

- you should read up, use the VFIO bind in tools, system devices, assign it to your VM and thats it

 

 

then the card should be free and so on ... thats why its not really the right place here as this plugin is for a different usecase

Hey. Yes, talking about just using it in a VM. Sorry for the confusion, I have not heard of VFIO bind until your post and assumed I needed the plugin for using a card. I was going off of Spaceinvader One's videos and dont believe he mentioned it in his VM videos (I could be wrong though). Will give this a try as soon as I get a chance. Thanks!

[Little709]

On 1/30/2022 at 10:06 PM, ich777 said:

Do you only tested the Plex web client or any native client too, completely forgot that sometimes the web client doesn't initialize the HW transcoding properly.

I'm back,

 

I did use the native client and the web client. 

 

But now, i've noticed that also Tdarr is not working any more. I get this error: 

 

Stream #0:0 -> #0:0 (h264 (h264_cuvid) -> hevc (hevc_nvenc))

Stream #0:1 -> #0:1 (copy)

Error while opening decoder for input stream #0:0 : Operation not permitted

 

I'm pretty positive that it's not permissions issues. I've ran the "safe permissions script".  I'm also pretty sure that it's just related to my GPU. 

 

I added a bug report by nvidia-bug-report.sh

 

I was wondering, is there any persistant data that is left behind on the system? Or are the drivers installed freshly after every reboot? If there is any persistant data, can i delete it all to start fresh?

nvidia-bug-report.log

Edited February 4, 2022 by Little709

[ich777]

7 hours ago, Little709 said:

I added a bug report by nvidia-bug-report.sh

Why? On unRAID it would be better that you post the Diagnostics.

 

7 hours ago, Little709 said:

I was wondering, is there any persistant data that is left behind on the system? Or are the drivers installed freshly after every reboot? If there is any persistant data, can i delete it all to start fresh?

You can delete the plugin, reboot, install it again, reboot and see if it is working.

 

The driver is installed on every reboot, yes. There is nothin gpeft on the system.

 

Would be also interesting if the containers that actually use the Nvidia driver are starting up after you've uninstalled the plugin and rebooted.

 

7 hours ago, Little709 said:

Error while opening decoder for input stream #0:0 : Operation not permitted

Have you tried Unmanic from @Josh.5 yet?

I think it's more related to the containers...

[Little709]

16 hours ago, ich777 said:

Why? On unRAID it would be better that you post the Diagnostics.

 

You can delete the plugin, reboot, install it again, reboot and see if it is working.

 

The driver is installed on every reboot, yes. There is nothin gpeft on the system.

 

Would be also interesting if the containers that actually use the Nvidia driver are starting up after you've uninstalled the plugin and rebooted.

 

Have you tried Unmanic from @Josh.5 yet?

I think it's more related to the containers...

No i haven't to be honest, but isn't it weird that both plex and tdarr are giving issues when using the GPU? Seems to me that it would be a broader problem than just the containers.

Btw, yesterday i remade all containers. I deleted my docker folder

 

I've added the diagnostics, right after I tried to transcode a file.

tower-diagnostics-20220205-1112.zip

[ich777]

2 hours ago, Little709 said:

No i haven't to be honest, but isn't it weird that both plex and tdarr are giving issues when using the GPU?

Did you already run the fix permissions on the folder where the transcoding directory is located?

 

Can you try just for testing purposes the Jellyfin container with HW transcoding?

 

2 hours ago, Little709 said:

Seems to me that it would be a broader problem than just the containers.

I'm really not sure since you've once reported that the container was working and HW transcoidng was working just fine a few posts back:

On 1/30/2022 at 9:59 PM, Little709 said:

And then i verify, and indeed, the error is gone....

 

Although the error does persist when using the windows client. Which is obviously quite weird.

 

Also, as said before on my system Emby, Jellyfin & Plex are working just fine with HW transcoding.

 

From the Diagnostics everything seems fine to me...

[Little709]

8 hours ago, ich777 said:

Did you already run the fix permissions on the folder where the transcoding directory is located?

 

Can you try just for testing purposes the Jellyfin container with HW transcoding?

 

I'm really not sure since you've once reported that the container was working and HW transcoidng was working just fine a few posts back:

 

Also, as said before on my system Emby, Jellyfin & Plex are working just fine with HW transcoding.

 

From the Diagnostics everything seems fine to me...

I'm sorry i'm giving some conflicting information.

 

I thought it was fixed, and felt real dumb. Some times the transcode would work but at the same time, some times it wouldn't. i thought it might have been something with the clients. But yesterday I see that tdarr is also failing when trying to actually transcode. 

 

But as the error in plex is a permission issue and tdarr has an "operation not permitted" i'm starting to suspect it's something with my actual GPU.

 

plex is using /tmp as a transcode folder. Tdarr is using a different folder "/mnt/user/tdarr-cache" 

 

Jellyfin is in fact is working.

 

I am really at a dead end. Even contemplating to try and use a new install of unraid to see if the problem persists

[ich777]

4 minutes ago, Little709 said:

I am really at a dead end. Even contemplating to try and use a new install of unraid to see if the problem persists

First of all, this will certainly not help because from the Diagnostics everything looks fine and a reinstall of unRAID will not much do I think.

 

5 minutes ago, Little709 said:

Jellyfin is in fact is working.

Please try it the next few days too if it works fine or not.

 

5 minutes ago, Little709 said:

But as the error in plex is a permission issue and tdarr has an "operation not permitted" i'm starting to suspect it's something with my actual GPU.

Is it maybe possible that some path is set up wrong and that the Docker image or something is running full and not space for the containers is left to start the transcode?

 

6 minutes ago, Little709 said:

But as the error in plex is a permission issue and tdarr has an "operation not permitted" i'm starting to suspect it's something with my actual GPU.

I don't think that your GPU is failing if Jellyfin just works fine.

Can you maybe try the container from @Josh.5 (Unmanic) and see if it is working fine with Nvidia hardware transcoding.

I've now heard from many people that they switched from whatever to Unmanic and they are all happy with it once it's configured (enough YouTube videos are out there how to set it up).

[Josh.5]

We have a discord community full of helpful people also available to help (myself included) https://unmanic.app/discord

[comet424]

i apparently have warning from nvida in my system logs

my system log says

Feb  7 08:19:38 Tower root: | WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING
Feb  7 08:19:38 Tower root: | WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING

 

tower-diagnostics-20220207-0840.zip

[Squid]

It's not an issue.  It's just part of the Installation messages where it tells you to not close the install window until it's finished.

[ich777]

57 minutes ago, comet424 said:

i apparently have warning from nvida in my system logs

my system log says

I would recommend that you search for the full message...

Have you deleted the folder /boot/config/plugins/nvidia-driver or do you have the plugin file manually placed on the USB Boot device or do you have maybe changed something in the settings.cfg for the plugin?

From what I see it seems that the first installation happened on boot on your server what is really suspicious to me...

 

EDIT: May I also ask why did you install the plugin? You have no Nvidia GPU installed...

[comet424]

ah ok..  and i had updated  my Unassigned devices plugin  and i did a reboot.. and got the warning warning warning  

 

and i didnt know you had to leave the check box  "text"  checked to see more.. i usually un check it.. so i can just see  warnings and errors..

 

i used to have a nvidia card in there... for passing VMs  but it burned out...    and figured i needed the driver for my card...  

but i 100% never noticed any improvements  using the unraid nvida driver when i had a nvida card in for my VMs

 

and if it doesnt help vms  when  do u need the driver for unraid itself?

 

and i dont know if i have the file manually on the usb  or not.. i just remember  installing the nvida  plugin from CA   more then a year or 2 ago.. and that was last time i touched it 

Edited February 8, 2022 by comet424