I recently noticed an issue when i upgraded to version 6.12.14 with my GTX 1070.

 

i get an error nvidia smi not found.  

 

I tried a few to reinstall the nvidia plugin. 

I tried to reseat the GPU.  The GPU lights up and  fans spins.

GPU is not listed in terminal when listing all the devices connected to the server (lspci) 

I tried downgrading to 6.12.13, but that did not help.

i tried to remove the contents of this folder /boot/config/plugins/nvidia-driver/packages/6.1.118/

 

i tried to set is as an open source driver.  after reboot i could see the gpu for a few min and then i got an different error.-  i dont have an screenshots of the other error.

 

 

 

does anyone have an suggestions?  i also added the diagnostics too 

 

Than you for your help

 

gormuck-diagnostics-20241128-0829.zip

4 hours ago, frothy said:

does anyone have an suggestions?  i also added the diagnostics too 

 

driver installation looking ok, but the card is "not there"

 

so i assume you already checked the seating, power supply, .. may take a look if something went loose,

also BIOS, above 4g decode and rbar activated if possible.

 

only concern i would see, many errors here ...

 

Nov 27 23:59:00 Gormuck kernel: I/O error, dev sr0, sector 49144 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2
Nov 27 23:59:00 Gormuck kernel: sr 0:0:0:1: [sr0] tag#0 unaligned transfer

 

but actually i cant tell if they could be related somehow ...

9 minutes ago, alturismo said:

driver installation looking ok, but the card is "not there"

 

so i assume you already checked the seating, power supply, .. may take a look if something went loose,

also BIOS, above 4g decode and rbar activated if possible.

 

only concern i would see, many errors here ...

 

Nov 27 23:59:00 Gormuck kernel: I/O error, dev sr0, sector 49144 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2
Nov 27 23:59:00 Gormuck kernel: sr 0:0:0:1: [sr0] tag#0 unaligned transfer

 

but actually i cant tell if they could be related somehow ...

The 1070 worked fine up until a few days ago.  I can try to test with a different NVIDIA graphics card.

 

is there anything i can do about the errors?

54 minutes ago, frothy said:

The 1070 worked fine up until a few days ago.  I can try to test with a different NVIDIA graphics card.

Have you yet tried to fully shutdown the server, pull the power cord from the wall, press the power and reset button a few times (to empty the caps), wait for at least a minute, plug the server back into the wall and try to boot?

 

Maybe something just hangs, if that also doesn't help the card might be defective...

The power pins were not seated properly in the GPU.  EVERYTHING WORKS.  Thank you for your help.  Much appreciated.

After upgrading to rc-1 I noticed kasm stopped working. Upon further investigation it seems the latest 565 driver does not create the requisite /sys/class/drm entries. Downgrading to 560 "new feature branch" fixes it. 

Unsure if this is an issue with the plugin, the nvidia driver, or rc-1, but posting here first since if nothing else I'll be redirected accordingly.

Edited December 4, 20241 yr by Spitko

2 hours ago, Spitko said:

Unsure if this is an issue with the plugin, the nvidia driver, or rc-1, but posting here first since if nothing else I'll be redirected accordingly.

Thank you for the report, I will look into this ASAP and report back!

3 hours ago, Spitko said:

After upgrading to rc-1 I noticed kasm stopped working.

Sorry but I can't reproduce this over here and everything seem to work totally fine:
 

 

 

Even transcoding works fine:

 

 

It would be super helpful if you could pull the Diagnostics next time when this happens so that I can take a deeper look why it was not working for you.

Hi @ich777,

 

I am having an issue with Jellyfin transcoding ffmpegs. I read through a few of the previous posts on this issue and the solutions there don't seem to be what I am looking for. 

 

Based on the versioning issue shown in the log I updated the driver, but for some reason this still seems to be an issue. Any direction you can point me in would be helpful. Thank!

 

Nvidia-smi screenshot as well:

error_log.txt

20 minutes ago, Prescott_2415 said:

I am having an issue with Jellyfin transcoding ffmpegs. I read through a few of the previous posts on this issue and the solutions there don't seem to be what I am looking for. 

Please post your Diagnostics.

 

Have you read the second post on how to pass through the GPU to a container?

What container (repository) are you using?

 

Are you on a really old version from Unraid or why are you on Nvidia Driver version: 545.29.06?

 

Please upgrade to a newer version from Unraid to get newer drivers since this seems to be your issue:

[h264_nvenc @ 0x5618f57a3040] Driver does not support the required nvenc API version. Required: 12.2 Found: 12.1

 

2 hours ago, ich777 said:

Please post your Diagnostics.

 

Have you read the second post on how to pass through the GPU to a container?

What container (repository) are you using?

 

Are you on a really old version from Unraid or why are you on Nvidia Driver version: 545.29.06?

 

Please upgrade to a newer version from Unraid to get newer drivers since this seems to be your issue:

[h264_nvenc @ 0x5618f57a3040] Driver does not support the required nvenc API version. Required: 12.2 Found: 12.1

 

 

I didn't realize I was so far behind on my Unraid version. I updated that and it looks like everything is working now. Thank you for your quick reply and help!

Hello.

I can't seem to know why the Nvidia Driver download failed. I'm on 7.0.0-beta.3 and tried to upgrade to 7.0.0-rc.1

Diagnostics pasted.

 

 

 

unraid-diagnostics-20241206-1048.zip

52 minutes ago, haco said:

I can't seem to know why the Nvidia Driver download failed. I'm on 7.0.0-beta.3 and tried to upgrade to 7.0.0-rc.1

It seems that your plugin is broken, please look at the second recommended post on top of this thread to resolve the issue, this is a indication that your plugin is broken:

 

However in your case I would recommend a little different route:

1. Remove the plugin
2. Reboot to upgrade
3. Install the plugin
4. Reboot again

I'm on Unraid version 6.9.2 and the latest available driver version is v515.43.04. I'm using my RTX2060 with my Frigate docker container, but it needs at least v530 in order to use TensorRT for object detection.

5 hours ago, Professor Oak said:

but it needs at least v530 in order to use TensorRT for object detection.

then you should consider to update unraid ...

 

5 hours ago, Professor Oak said:

I'm on Unraid version 6.9.2

as you are far behind ... there wont be any driver updates on "old" builds, way too much effort to compile all driver packages vs. all kernel versions and so on ...

Hello, I'm facing the issue with "NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

I have run the server on Unraid 6.12.13 for a long time and I have a Quadro P400 on my Dell R720 server. I keep to use the display card for jellyfin transcode and it works well, until I decide to upgrade the server versionto 6.12.14. I also upgrade the plugin version from 2024.12.08 to 2024.07.10. I can't remember which Nvidia driver I use before but I tried to download the newest driver.

After restart, the plugin shows the error as above. I'm not quite familiar with the nvidia driver. How could I investigate the error?

The diagnostics file is attached.

tower-diagnostics-20241211-2028.zip

1 hour ago, Mr.win7 said:

After restart, the plugin shows the error as above. I'm not quite familiar with the nvidia driver. How could I investigate the error?

 

the card is not listed in your pci devices ... so driver plugin cant see a card and then ...

 

may take a closer look at

 

1/ seating

2/ power supply

3/ last but not least ... still working ?

 

and you didnt change anything on hardware ... ?

16 minutes ago, alturismo said:

the card is not listed in your pci devices ... so driver plugin cant see a card and then ...

 

may take a closer look at

 

1/ seating

2/ power supply

3/ last but not least ... still working ?

 

and you didnt change anything on hardware ... ?

Nothing change on hardware. Thank you for your help. I will have a try on unplug and plug it again.

1 hour ago, Mr.win7 said:

Quadro P400

Sadly enough I saw a few cards dying from that type alongside with some P2000 cards.

 

However as @alturismo already pointed out the card is not visible in your System.

Concerning the post from ocean_breeze Posted April 24, 2023.
I do have the same problem, booting Unraid on GUI mode gives after loading and being ready to load the GUI a blank screen.
I'm trying desperately to find a solution to this issue.
I know that it is caused by the Nvidia Driver, since once I uninstall it, the loading of the GUI works as it should.
As soon as I reinstall the driver, I get a blank screen when booting in GUI mode.
In text mode, no problem at all.

2 hours ago, LordVader said:

As soon as I reinstall the driver, I get a blank screen when booting in GUI mode.

Without Diagnostics I really can‘t help.

Sorry about that.
Here it is.
I'm a bit green on the unraid / unix side.

maverick-diagnostics-20241211-1509.zip

1 hour ago, LordVader said:

Sorry about that.

Are you really sure that your Nvidia GPU is the primary output and not the AMD one?

 

However if that is not the case please issue this command once from a Unraid terminal:

sed -i "/disable_xconfig=/c\disable_xconfig=true" /boot/config/plugins/nvidia-driver/settings.cfg

and reboot your server.

Thank you for your help.
It worked, the GUI is now showing after the completition of the loading process.

Fantastic.
Out of curiosity.
What that command dit?

Just now, LordVader said:

What that command dit?

The command sets a flag so that the plugin doesn't run nvidia-xconf on boot.

This will disable accelerated GUI output but will allow you to use multiple cards in one system.

 

Running Unraid in GUI mode is nowadays a edge case anyways and running it with multiple GPUs from different brands causes this issue.