[Plugin] Nvidia-Driver

[ich777]

11 minutes ago, Zotarios said:

I wouldn't know how to do so, because the webui is not responsible. Maybe I can try to boot the server with the gui image?

Oh, did you say that in your post above?

 

But I don‘t think that the Nvidia Driver is actullay the issue because it is just complaining that it can‘t find the drivers.

 

Are you sire that the network cable is plugged in correctly and that the server is actually responding to a ping on the shown IP address?

 

11 minutes ago, Zotarios said:

Same problem... how could I get them?

Log in an type in:

diagnostics

 

Then you‘ll find the Diagnostics on the USB Boot device if you plug it in to your PC.

 

However as said, I think that you actually run into a network error, these messages won‘t prevent the GUI from showing up.

[KingHawk]

I'm having some trouble getting my new RTX 4070 Ti Super working on unraid. When I install the new gpu the Nvidia Driver plugin shows [Unknown Error] under 'Installed GPU(s):'
The latest version of the NVIDIA driver (in the plugin) is installed and this worked perfectly fine with the old gpu GTX 1050 Ti.
Diagnostics attached, hope it helps.

jj-silverstone-diagnostics-20240207-1346.zip

 

[ich777]

1 minute ago, KingHawk said:

The latest version of the NVIDIA driver (in the plugin) is installed and this worked perfectly fine with the old gpu GTX 1050 Ti.

No support with this:

Feb  7 13:36:48 JJ-SILVERSTONE emhttpd: /usr/local/emhttp/plugins/user.scripts/backgroundScript.sh "/tmp/user.scripts/tmpScripts/unlock nvidia/script" >/dev/null 2>&1

 

You know that Nvidia now upped the simultaneous transcodes to 8 silently with the newest driver that you can't install because you are still on Unraid 6.12.3 where Unraid 6.12.6 is already available.

 

The latest driver on Unraid 6.12.3 is not the latest driver which is available on Unraid 6.12.6 which you will need for the new Super series.

[KingHawk]

15 hours ago, ich777 said:

No support with this:

Feb  7 13:36:48 JJ-SILVERSTONE emhttpd: /usr/local/emhttp/plugins/user.scripts/backgroundScript.sh "/tmp/user.scripts/tmpScripts/unlock nvidia/script" >/dev/null 2>&1

 

You know that Nvidia now upped the simultaneous transcodes to 8 silently with the newest driver that you can't install because you are still on Unraid 6.12.3 where Unraid 6.12.6 is already available.

 

The latest driver on Unraid 6.12.3 is not the latest driver which is available on Unraid 6.12.6 which you will need for the new Super series.

I didn't know about that thanks! I will try this tomorrow morning.

EDIT: Successfully update everything and disabled script. New gpu successfully installed.

Edited February 8 by KingHawk

[Kleinish888]

looking to add a tesla p100 to my unraid server. I have the data center driver. how would I go about compiling it for unraid? The prices if a p100 is low as of late just wanted to see if I could get it to work.

Edited February 7 by Kleinish888
added picture

[ich777]

8 minutes ago, Kleinish888 said:

The prices if a p100 is low as of late just wanted to see if I could get it to work.

What's the exact issue?

 

Is it the second card? It is working as far as I can see but without Diagnostics I can't say much.

[Kleinish888]

I was looking to replace the p2000 in my server and use the p100 for transcoding on plex. With the current driver I get a driver crash likely because of the incompatibility with the card. I was figuring that if I compile the vgpu driver that I use on my proxmox server to work with unraid. I could use the p100 for this task

[ich777]

2 minutes ago, Kleinish888 said:

I was looking to replace the p2000 in my server and use the p100 for transcoding on plex. With the current driver I get a driver crash likely because of the incompatibility with the card. I was figuring that if I compile the vgpu driver that I use on my proxmox server to work with unraid. I could use the p100 for this task

Can you please post your Diagnostics?

The Tesla P100 series should work.

 

I have no idea what is going on on your system.

[Kleinish888]

here is the file. I have been doing some testing it seem to be working. however I am noticing when transcoding more then one 4K DoVi/HDR10 (HEVC Main 10) (hw). The movie will not load

Edited February 7 by Kleinish888
removing my data

[ich777]

50 minutes ago, Kleinish888 said:

I have been doing some testing it seem to be working.

I see no indication in your Diagnostics that it shouldn't work.

 

50 minutes ago, Kleinish888 said:

however I am noticing when transcoding more then one 4K DoVi/HDR10 (HEVC Main 10) (hw).

And was it working with the card that you used before?

This could also be an issue with the container itself that you are using.

[Kleinish888]

lscr.io/linuxserver/plex:latest is the docker container

[ich777]

34 minutes ago, Kleinish888 said:

lscr.io/linuxserver/plex:latest is the docker container

I really don‘t know what the issue can be, can you just to double check the official Jellyfin container and see if it is working there?

 

Was DV working before on your old card?

[Kleinish888]

I just made the change this morning. lets let it buck for a few days, to see if any of my user notice anything out of the ordinary. thank your for your help I will add the jellyfin container and see if I notice any differences 

[aeryk]

My 3080 Ti doesn't show up to use for docker. It allows it to passthrough to VMs, but GPU Statistics doesn't show any GPUs. I don't have any VMs running.

[ich777]

22 minutes ago, aeryk said:

My 3080 Ti doesn't show up to use for docker. It allows it to passthrough to VMs, but GPU Statistics doesn't show any GPUs. I don't have any VMs running.

I already saw your posts in the GPU Statistics plugin support thread.

Does it work in your container or does it simply not show up?

 

As always without attached Diagnostics I can't help.

[aeryk]

27 minutes ago, ich777 said:

I already saw your posts in the GPU Statistics plugin support thread.

Does it work in your container or does it simply not show up?

 

As always without attached Diagnostics I can't help.

It doesn't show up or work inside containers

 

I spotted something in the logs. DM'd them to you ❤️

 

Feb  7 17:42:45 Tower kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 242
Feb  7 17:42:45 Tower kernel: NVRM: The NVIDIA probe routine was not called for 1 device(s).
Feb  7 17:42:45 Tower kernel: NVRM: This can occur when a driver such as: 
Feb  7 17:42:45 Tower kernel: NVRM: nouveau, rivafb, nvidiafb or rivatv 
Feb  7 17:42:45 Tower kernel: NVRM: was loaded and obtained ownership of the NVIDIA device(s).
Feb  7 17:42:45 Tower kernel: NVRM: Try unloading the conflicting kernel module (and/or
Feb  7 17:42:45 Tower kernel: NVRM: reconfigure your kernel without the conflicting
Feb  7 17:42:45 Tower kernel: NVRM: driver(s)), then try loading the NVIDIA kernel module
Feb  7 17:42:45 Tower kernel: NVRM: again.
Feb  7 17:42:45 Tower kernel: NVRM: No NVIDIA devices probed.

 

 

[ich777]

4 minutes ago, aeryk said:

It doesn't show up or work inside containers

Sorry but you are still on 6.11.5, why?

Please upgrade to 6.12.6 and install the latest Nvidia drivers there since you are on driver version: 535.54.03 and the latest one is 550.40.07 (...no I won't compile the new drivers for such an old Unraid version).

 

However the problems you are experiencing can be caused because you are using it for a VM and Docker containers, it will of course work in most situations but not always.

 

Please upgrade to Unraid 6.12.6 and see if the issue persists.

[aeryk]

16 hours ago, ich777 said:

Sorry but you are still on 6.11.5, why?

Please upgrade to 6.12.6 and install the latest Nvidia drivers there since you are on driver version: 535.54.03 and the latest one is 550.40.07 (...no I won't compile the new drivers for such an old Unraid version).

 

However the problems you are experiencing can be caused because you are using it for a VM and Docker containers, it will of course work in most situations but not always.

 

Please upgrade to Unraid 6.12.6 and see if the issue persists.

Oh the reason is I don't have access to the box until next month and I'm worried if I upgrade it might break entirely. It's okay if some part is broken, but if it doesn't boot, that's a problem. I know it's not your responsibility, but does anybody know the likelihood of an update failing? (I'm worried because updating Plex with a wrong argument created an "orphaned docker" so who knows).

[ich777]

4 hours ago, aeryk said:

It's okay if some part is broken, but if it doesn't boot, that's a problem. I know it's not your responsibility, but does anybody know the likelihood of an update failing?

I don't think that a normal upgrade should prevent the boot, but you'll never know.

 

4 hours ago, aeryk said:

I'm worried because updating Plex with a wrong argument created an "orphaned docker" so who knows

...what is the exact issue with that? This is just the Docker image for Plex because it can't start, if you fix the issue in the template and recreat the container the orphaned image will be used again for Plex (you could also delete a orphaned image and it will be pulled again from DockerHub if needed).

[pashashocky]

Heya, I'm wondering what is your guidance on using this plugin with GPU being occasionally passed into a VM?

 

I couldn't find any information apart from don't download if you are planning to pass through to a VM, but I've read in other forum posts that I could pass it through to a VM when it's not being used by a docker container?

 

Plex can see my GPU right now, but I've selected the iGPU for transcoding.. does it mean it's also being used by docker and I can't pass it through to a VM? 

 

Feels like I have to constantly install and remove this plugin if I want to ever pass through to the VM, but is there any way to do it without constant server reboots?

Edited February 14 by pashashocky
more content

[alturismo]

6 hours ago, pashashocky said:

Heya, I'm wondering what is your guidance on using this plugin with GPU being occasionally passed into a VM?

 

Page 1 ... readme ... Post 1 the description and notes about VM Usage and Docker Usage ...

[alturismo]

6 hours ago, pashashocky said:

but I've read in other forum posts that I could pass it through to a VM when it's not being used by a docker container?

yes, thats correct

 

6 hours ago, pashashocky said:

but I've selected the iGPU for transcoding.. does it mean it's also being used by docker and I can't pass it through to a VM? 

if you have a iGPU, then VFIO bind your Nvidia card so its gone from the Host ... then you have no issues passing your NV card to a VM while using the iGPU for transcoding.

 

thats the most common setup ... otherwise its always a "either or" ... Docker OR VM ... and can likelycrash your Server.

[pashashocky]

10 hours ago, alturismo said:

thats the most common setup ... otherwise its always a "either or" ... Docker OR VM ... and can likelycrash your Server.

 

Yeah, so wha I'm trying to figure out is how do I go around the "will likely crash your server" - is it random or does that only happen if the GPU is being used elsewhere? binding to vfio is suboptimal because the GPU eats power all the time then.

[alturismo]

1 minute ago, pashashocky said:

Yeah, so wha I'm trying to figure out is how do I go around the "will likely crash your server" - is it random or does that only happen if the GPU is being used elsewhere?

exactly ... if another app tries to access the device while "in use" ...

 

4 minutes ago, pashashocky said:

binding to vfio is suboptimal because the GPU eats power all the time then.

your decision, if your Server crashes ou know why

 

actually, i use it here like this with 2 VM's and my dGPU's are not bound to VFiO ... all Apps (Dockers) where only pointed to /dev/dri to use the iGPU and i personally had no issues, but you are warned and know why ... in case ...

 

 

and i use nvidia-persistenced to keep them in sleep while not in use (VMs off, as far as it goes ...)

[vedadechiee]

having an issue with my official plex docker unraid 6.12.6 nvidia driver installed and followed the instructions. plex shows the gpu but is not using it for transcoding. anything im doing wrong? i also cant update the nvidia drivers to a different version it says "nothing to do driver 535.146.02 already downloaded" not sure if thats effecting anything.

homel1-syslog-20240216-0458.zip