[Plugin] Nvidia-Driver

[AndrewClaus]

14 hours ago, ich777 said:

You have a lot of:

Oct 20 18:47:00 Tower kernel: NVRM: GPU 0000:00:10.0: RmInitAdapter failed! (0x25:0x65:1411)
Oct 20 18:47:00 Tower kernel: NVRM: GPU 0000:00:10.0: rm_init_adapter failed, device minor number 0

 

in your Syslog, is this the first time you are trying to install the Nvidia Driver?

 

Please make sure that you've disabled C-States in your BIOS and that you enable Above 4G-Decoding and Resizable BAR Support if you have these options in your BIOS.

Thanks, those BIOS settings worked! My GPU is now being read by the Nvidia Driver:

However when having transcoding enabled on my Plex, video fails to play with an Error code: s1003. The docker container cannot be stopped via GUI (pictured below) or killed via terminal. I had to force stop my Unraid VM.

With all the part 2 configuration but transcoding disabled within the Plex app the videos play fine, it's only when I enable the acceleration settings:

Here is my part 2 configuration:

 

Attached is my diagnostics. Any idea?

tower-diagnostics-20221021-1456.zip

 

 

Edited October 21, 2022 by AndrewClaus

[ich777]

6 minutes ago, AndrewClaus said:

The docker container cannot be stopped via GUI or killed via terminal. I had to force stop my Unraid VM.

If you are virtualizing Unraid I really can‘t help because I use it on bare metal and one wrong setting on the Host can prevent the card from working when virtualizing it.

 

Maybe ask on the Proxnox forums or over in the Virtualisazion Sub Forums like @trurl pointed already out.

[AndrewClaus]

21 minutes ago, ich777 said:

Do you use one card for a VM and transcoding? If yes, don‘t do that!

 

Or are you virtualizing Unraid?

If you are virtualizing Unraid I really can‘t help because I use it on bare metal and one wrong setting on the Host can prevent the card from working.

I'm running Unraid as a VM within Proxmox and passing the GPU through to Unraid.

 

I'll revisit the passthrough instructions again and ask elsewhere. Worst case I run Unraid bare metal too, it supports VMs too but I need to compare it's limitations. Thanks

Edited October 21, 2022 by AndrewClaus

[ich777]

3 hours ago, AndrewClaus said:

Bare metal Unraid worked immediately! Do you recommend something like

Please read the first post, the red text on top.

[abasedepoppoppoppop]

Is there an issue with the nvidia plugin. I cannot donwload legacy drivers v470.141.03. I am using an old GT 730

[ich777]

1 hour ago, abasedepoppoppoppop said:

Is there an issue with the nvidia plugin. I cannot donwload legacy drivers v470.141.03. I am using an old GT 730

Please post your Diagnostics, without them I can‘t tell much.

[JC2020]

Hello,

 

Since I upgraded to 6.11 I have been getting an issue where one of my GPU's disappear from the Nvidia DIag plugin.  I have found in the logs this:
Oct 24 10:04:27 Tower kernel: NVRM: GPU 0000:04:00.0: RmInitAdapter failed! (0x23:0xffff:1365)
Oct 24 10:04:27 Tower kernel: NVRM: GPU 0000:04:00.0: rm_init_adapter failed, device minor number 0
Oct 24 10:04:27 Tower kernel: NVRM: GPU 0000:04:00.0: RmInitAdapter failed! (0x23:0xffff:1365)
Oct 24 10:04:27 Tower kernel: NVRM: GPU 0000:04:00.0: rm_init_adapter failed, device minor number 0
Oct 24 10:04:28 Tower kernel: NVRM: GPU 0000:04:00.0: RmInitAdapter failed! (0x23:0xffff:1365)
Oct 24 10:04:28 Tower kernel: NVRM: GPU 0000:04:00.0: rm_init_adapter failed, device minor number 0
Oct 24 10:04:28 Tower kernel: NVRM: GPU 0000:04:00.0: RmInitAdapter failed! (0x23:0xffff:1365)
Oct 24 10:04:28 Tower kernel: NVRM: GPU 0000:04:00.0: rm_init_adapter failed, device minor number 0

 

Any help would be appreciated.

 

Edited October 24, 2022 by JC2020

[ich777]

31 minutes ago, JC2020 said:

Any help would be appreciated.

Please upload your Diagnostics.

[JC2020]

tower-diagnostics-20221024-1208.zip

[ich777]

56 minutes ago, JC2020 said:

tower-diagnostics-20221024-1208.zip 192.7 kB · 0 downloads

Seems like you have an issue with your mining container:

Oct 24 02:01:32 Tower kernel: miner1[3117]: segfault at 21 ip 0000000000467ce4 sp 0000147641dfd740 error 4 in t-rex[400000+19fa000]
Oct 24 02:01:32 Tower kernel: Code: df e8 30 7b 4f 00 48 83 7b 58 00 0f 84 8d 00 00 00 48 8b 43 48 0f b6 50 21 31 c0 84 d2 74 54 48 83 c4 18 5b 5d c3 48 8b 43 48 <0f> b6 40 21 84 c0 0f 94 c0 48 83 c4 18 5b 5d c3 0f 1f 40 00 48 8b

 

Are you sure your card is working properly, was this card working before?

[JC2020]

Yes, when I restart unraid it works for a day to a few days then disappears. 

[ich777]

2 minutes ago, JC2020 said:

Yes, when I restart unraid it works for a day to a few days then disappears. 

Do you have Above 4G Decoding and Resizable BAR Support in your BIOS?

 

What you can also try is to disable ASPM in your BIOS and also boot with CSM (Legacy) instead of UEFI.

[JC2020]

So I have a ASRock EP2C602 I'll have to check since I haven't been in BIOS in awhile and it's remote to me.  Just odd that this started happening recently as nothing has changed in my setup except unraid updates.

 

[ich777]

16 minutes ago, JC2020 said:

Just odd that this started happening recently as nothing has changed in my setup except unraid updates.

It is also possible that the miner crashes your graphics card and that it drops off the bus and it isn't able to recover from that.

[JC2020]

Ah possibly.  Is there anyway other than a reboot to try and restart the GPU?  I will shutoff the miner and give it a go to see if I lose the GPU again.

[ich777]

5 hours ago, JC2020 said:

Is there anyway other than a reboot to try and restart the GPU?

Yes and no, it is rather complicated when the card falls from the bus and how hard it crashed but is not always guaranteed that it is working.

 

Please also make sure that you enable, disable the options that I've mentioned already in the BIOS.

[ich777]

3 hours ago, PneuMatix said:

Any help would be appreciated.

I think the first thing that you should really do is buy a valid license, after you've did that I and @SpencerJ are happy to help.

[SpencerJ]

3 hours ago, PneuMatix said:

Hello, I'm new to UNRAID and am currently setting up JellyFin with 1650 super. I downloaded the Nvidia Driver, but it appear "NVIDIA-SMI failed due to a communication failure with the NVIDIA driver. Check that the most recent NVIDIA driver is installed and running." Tried going through step by step a few times and searching but through the issue but maybe I'm missing something.

I've confirmed that the card works in another machine and that it appears in my server's hardware list.

I tried different versions of the graphic card driver, but the message remained the same.

Any help would be appreciated.

tower-diagnostics.zip 59.03 kB · 3 downloads

 

https://unraid.net/buy-genuine-license

[Findthelorax]

Can't start any dockers with --runtime=nvidia

they all say "Bad Parameter"

 

any advice on what to look at?

docker run
  -d
  --name='frigate'
  --net='custom'
  --privileged=true
  -e TZ="America/New_York"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="##"
  -e HOST_CONTAINERNAME="frigate"
  -e 'FRIGATE_RTSP_PASSWORD'='##'
  -e 'NVIDIA_VISIBLE_DEVICES'='GPU-20e89059-##'
  -e 'NVIDIA_DRIVER_CAPABILITIES'='all'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:5000]'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/yayitazale/unraid-templates/main/frigate.png'
  -p '5000:5000/tcp'
  -p '1935:1935/tcp'
  -v '/mnt/cachessd/appdata/frigate':'/config':'rw'
  -v '/mnt/user/Documents/Media/frigate':'/media/frigate':'rw'
  -v '/tmp/frigate':'/tmp/cache':'rw'
  -v '/etc/localtime':'/etc/localtime':'rw'
  --device='/dev/usb'
  --shm-size=5G
  --restart unless-stopped
  --runtime=nvidia 'blakeblackshear/frigate:stable-amd64'

d1ad2297da28b5f58cb0c0c08aca28f0f1cec9ffb0395a0f122eb1d2551bfb58
docker: Error response from daemon: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #1: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
nvidia-container-cli: initialization error: open failed: /proc/sys/kernel/overflowuid: permission denied: unknown.

The command failed.

 

diagnostics-20221030-0832.zip

Edited October 30, 2022 by Findthelorax

[ich777]

1 hour ago, Findthelorax said:

Can't start any dockers with --runtime=nvidia

Please post your Diagnostics.

[UBS]

On 10/30/2022 at 3:43 PM, ich777 said:

Please post your Diagnostics.

I have the same issue, and it started all of a sudden.

may be a recent plug-in update made it, as it was originally working on 6.11.1

 

 

unraid-diagnostics-20221103-0008.zip

[ich777]

7 hours ago, UBS said:

I have the same issue, and it started all of a sudden.

Do you run the script from SpaceInvaderOne?

Please remove that script, reboot and see if it's the same.

 

If that doesn't help, please go in the container template from a affected container change something and change it back so that you can press the Apply button and see if anything changes after pressing Apply.

[ich777]

58 minutes ago, gamertaboo said:

Thank you for any help I really appreciate it usually I can figure this stuff out but this has me stumped.

Please make sure to enable Resizable BAR Support and Above 4G Decoding in your BIOS.

Also please try to boot with Legacy (CSM) Mode instead of UEFI.

 

Do you maybe did a BIOS update in between too? It seems like that something is set wrong in your BIOS:

Nov  2 18:30:57 Plex2 kernel: NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
Nov  2 18:30:57 Plex2 kernel: NVRM: BAR0 is 0M @ 0x0 (PCI:0000:2d:00.0)
Nov  2 18:30:57 Plex2 kernel: nvidia: probe of 0000:2d:00.0 failed with error -1

 

Since you are using a AMD CPU please also try to disable C-States.

[gamertaboo]

8 hours ago, ich777 said:

Please make sure to enable Resizable BAR Support and Above 4G Decoding in your BIOS.

Also please try to boot with Legacy (CSM) Mode instead of UEFI.

 

Do you maybe did a BIOS update in between too? It seems like that something is set wrong in your BIOS:

Nov  2 18:30:57 Plex2 kernel: NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
Nov  2 18:30:57 Plex2 kernel: NVRM: BAR0 is 0M @ 0x0 (PCI:0000:2d:00.0)
Nov  2 18:30:57 Plex2 kernel: nvidia: probe of 0000:2d:00.0 failed with error -1

 

Since you are using a AMD CPU please also try to disable C-States.

Hello,

I cleared CMOS and set every setting you asked me to still no luck. Thanks

tower-diagnostics-20221103-0903.zip

[ich777]

1 hour ago, gamertaboo said:

I cleared CMOS and set every setting you asked me to still no luck. Thanks

Are you really sure that you have enabled Above 4G Decoding and Resizable BAR Support?

Please double check if you have any other options for BAR or something like 64bit support in the BIOS.

 

Please also try to disable to onbaord Aspeed AST graphics if possible to see if this makes any difference.

 

I'm pretty certain that a setting in the BIOS is wrong, please also update your BIOS since you are on version P1.20 where P1.50 is available.