@ich777

- In the Bios I have only the Option for Resize Bar "disable" or "auto", and it sits on auto.

- "Above 4G Decoding" is ON.

- I thought I have disabled the c-states, but those were actually enabled... Corrected that now.

- PSU is 550 watt from the original build of this PC. Even with the GPU, the whole thing should not need more then 220 watts. So plenty wattage

- Pressing "power" 2 times didn't work  As I dont have speakers in this Case (Jonsbo V4), at least I didn't find cables for the case speaker, I waited at the case, if anything happens... unfortunatly nothing happend. If I do a shut down or restart from the GUI everything is fine.

 

Yes, I did not unplugg the card inbetween the several installation processes. And yes, the plugin was installed while I generated the diagnostics. I thought this would be better for the informations. But I can unplug the Card, uninstall the plugin and then get a new diagnostics if that would be helpful.

 

I now have some minutes and will do bios update, control all bios settings and check if anything happens in the GUI.

1 hour ago, BastiKA84 said:

I thought this would be better for the informations. But I can unplug the Card, uninstall the plugin and then get a new diagnostics if that would be helpful.

Try to set up a remote syslogserver or connect a monitor so that you can actually see the output from the console when it crashes, take a picture and post it here, that would be super helpful.

 

But as said above this seems like some kind of hardware compatibility issue and is maybe solved with a BIOS update.

 

1 hour ago, BastiKA84 said:

I now have some minutes and will do bios update, control all bios settings and check if anything happens in the GUI.

Please don't forget to disable C-States again after the BIOS update and validate all other settings.

Bios Update is done. All settings should be correct:

 

 

 

Starting the server and entering the GUI I stopped all dockers and installed the plugin again. The opening windows states, everything should be fine:

 

But unfortunatly, I still can not enter the plugin... The GUI just stoppes working.

Here, 5 minutes after trying to enter the plugin:

 

As the server is still connected to a display, I can see inputs from the keyboard, which is also still connected.

-------------------------------

Wow... Whats this?

As I tried starting the plugin while typing here, I did not do a hard reset as I was looking up how to do a SSH call (yes, I am that kind of a noob). I recognized, that the unraid tab in my browser was not loading anymore. It showed the page with all my installed plugins. I now pressed the plugin button again, and I am in the plugin... 

 

Seems running now. Thank you @ich777 for you support  Now I will try to add the Card to jellyfin for transcoding. So maybe I will be back in some minutes 😅

29 minutes ago, BastiKA84 said:

So maybe I will be back in some minutes 😅

Can you please pull Diagnostics?

 

I assume that this is some kind of hardware issue…

45 minutes ago, ich777 said:

Can you please pull Diagnostics?

 

I assume that this is some kind of hardware issue…

Yes sir! Here are the diagnostics.

 

Edit:

Transcoding on the T400 is working. As I am playing a video in jellyfin, the gpu statistics plugin shows activity on the card

 

unraidserver-diagnostics-20240422-1942.zip

Edited April 22, 20242 yr by BastiKA84

Hi All, 

 

I've upgraded my old P400 with a T4 but the new card is not recognized. 

 

Apr 23 19:08:28 littleboy kernel: NVRM: GPU at PCI:0000:af:00: GPU-36d51216-544d-71c6-0604-11d08f217cd0
Apr 23 19:08:28 littleboy kernel: NVRM: Xid (PCI:0000:af:00): 140, pid='<unknown>', name=<unknown>, An uncorrectable ECC error detected (possible firmware handling failure) DRAM:-1840691974, LTC:0, MMU:0, PCIE:0
Apr 23 19:08:28 littleboy kernel: NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x62:0x40:2523)
Apr 23 19:08:28 littleboy kernel: NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0
Apr 23 19:08:28 littleboy kernel: [drm:nv_drm_load [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x0000af00] Failed to allocate NvKmsKapiDevice
Apr 23 19:08:28 littleboy kernel: [drm:nv_drm_register_drm_device [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x0000af00] Failed to register device

  [...]

Apr 23 19:12:10 littleboy kernel: NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x62:0x40:2523)
Apr 23 19:12:10 littleboy kernel: NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0
Apr 23 19:12:10 littleboy kernel: nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
Apr 23 19:12:10 littleboy kernel: nvidia-uvm: Loaded the UVM driver, major device number 235.
Apr 23 19:12:10 littleboy kernel: NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x62:0x40:2523)
Apr 23 19:12:10 littleboy kernel: NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0

running nvidia-smi command I have this output: 

 

# nvidia-smi
No devices were found

 

# lsmod | grep nvidia
nvidia_uvm           4644864  0
nvidia_drm             90112  0
nvidia_modeset       1347584  1 nvidia_drm
nvidia              54116352  2 nvidia_uvm,nvidia_modeset
video                  61440  1 nvidia_modeset
drm_kms_helper        167936  4 mgag200,nvidia_drm
drm                   499712  6 drm_kms_helper,drm_shmem_helper,nvidia,mgag200,nvidia_drm
backlight              20480  3 video,drm,nvidia_modeset
i2c_core               86016  9 drm_kms_helper,i2c_algo_bit,igb,nvidia,mgag200,i2c_smbus,i2c_i801,ipmi_ssif,drm

 

I deleted the GPU Stats plugin, reinstalled drivers, and rebooted a couple of times, but the card is not recognized.

 

 

I read a bit around the forum and I have:

 

The GPU is also recognized in BIOS/iDrac

 

 

Is it possible to fix? Could be the GPU broken? 

 

 

 

littleboy-diagnostics-20240417-1506.zip

Edited April 23, 20242 yr by skler

2 hours ago, skler said:

I've upgraded my old P400 with a T4 but the new card is not recognized. 

This is a really bad sign:

Apr 23 19:08:28 littleboy kernel: NVRM: Xid (PCI:0000:af:00): 140, pid='<unknown>', name=<unknown>, An uncorrectable ECC error detected (possible firmware handling failure) DRAM:-1840691974, LTC:0, MMU:0, PCIE:0

 

It might be that your card is defective or your Firmware/BIOS is simply not compatible with your Nvidia card.

 

If possible please try to driver version: 550.67 in the plugin, reboot and see if that helps.

Can you maybe also try another slot in your Server?

 

It is also possible that the GPU is not compatible with your server, this would not be the first time that I see a card because it's not Dell, HP,... certified.

 

I assume auxiliary power is also connected properly?

 

For testing purposes I would also recommend that you remove the file: /boot/config/modprobe.d/nvidia.conf

Just a information because I saw it, you should do it like that if you want to use more options:

options nvidia-drm modeset=1 fbdev=1

 

7 minutes ago, ich777 said:

This is a really bad sign:

Apr 23 19:08:28 littleboy kernel: NVRM: Xid (PCI:0000:af:00): 140, pid='<unknown>', name=<unknown>, An uncorrectable ECC error detected (possible firmware handling failure) DRAM:-1840691974, LTC:0, MMU:0, PCIE:0

 

It might be that your card is defective or your Firmware/BIOS is simply not compatible with your Nvidia card.

 

If possible please try to driver version: 550.67 in the plugin, reboot and see if that helps.

 

Done, Diagnostic attached (without nvidia.conf params)

 

7 minutes ago, ich777 said:

Can you maybe also try another slot in your Server?

 

is the only pcie x16

 

7 minutes ago, ich777 said:

 

It is also possible that the GPU is not compatible with your server, this would not be the first time that I see a card because it's not Dell, HP,... certified.

 

it should be compatible.. 

 

https://docs.nvidia.com/certification-programs/nvidia-certified-systems/index.html

 

 

 

7 minutes ago, ich777 said:

 

I assume auxiliary power is also connected properly?

 

the T4 don't have one 

 

7 minutes ago, ich777 said:

 

For testing purposes I would also recommend that you remove the file: /boot/config/modprobe.d/nvidia.conf

 

done

 

7 minutes ago, ich777 said:

Just a information because I saw it, you should do it like that if you want to use more options:

options nvidia-drm modeset=1 fbdev=1

 

 

oki thanks 

 

still no device found 

littleboy-diagnostics-20240423-2225.zip

I've tried opensource drivers too

 

[  256.084894] nvidia-uvm: Loaded the UVM driver, major device number 236.
[  257.340047] NVRM: kgspInitRm_IMPL: unexpected WPR2 already up, cannot proceed with booting gsp
[  257.340054] NVRM: kgspInitRm_IMPL: (the GPU is likely in a bad state and may need to be reset)
[  257.340060] NVRM: RmInitAdapter: Cannot initialize GSP firmware RM
[  257.343106] NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x62:0x40:1784)
[  257.345186] NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0

 

do you think that the GPU could be broken? 

 

littleboy-diagnostics-20240423-2250.zip

 

9 hours ago, skler said:

Done, Diagnostic attached (without nvidia.conf params)

Did you already try it with the driver version that I recommended above?

 

9 hours ago, skler said:

it should be compatible.. 

But do you maybe have to buy a license to make use of it? Such Servers can be horrible when it comes to hardware support.

I can't help much with real server stuff because I only use consumer hardware on my system.

 

9 hours ago, skler said:

I've tried opensource drivers too

Thank you, since this is a real Datacenter card you don't have to use the option in nvidia.conf like pointed out on the plugin page but it shouldn't make a real difference.

 

9 hours ago, skler said:

do you think that the GPU could be broken? 

Maybe yes but it is really hard to tell, are you also sure that the PCIe x16 slot can deliver enough power, I think someone with a Dell PowerEdge had an issue where the PCIe slot doesn't delivered enough power because of a riser card.

 

Can you maybe test the card in another system and install the Nvidia driver?

56 minutes ago, ich777 said:

Can you maybe test the card in another system and install the Nvidia driver?

 

I can do this tomorrow, restoring some old hw. btw thanks you in the meantime 

Hi everyone.  I'm having a similar issue with the 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.

 

I've got a Tesla P4... and Oddly enough it was working fine.. But it was getting really hot and I saw this error...... So i took it out, got a fan adapter for it that i'm powering w/a different power supply, and I still have this SMI issue.

 

I've attached diagnostics, but I hope it doesn't suggest there is an issue with the actual card.

 

I've reinsalled the driver, I'm moved the card to a different PCIE  and still not able to get passed this error.

 

Any thoughts or suggestions would be greatly appreciated.

odyssey-diagnostics-20240425-1029.zip

24 minutes ago, mew0717 said:

I've reinsalled the driver, I'm moved the card to a different PCIE  and still not able to get passed this error.

Sorry but I don't see a Nvidia GPU on your PCIe bus.

 

Are you sure the card isn't defective. Please try to reseat the card and make sure that it plugged in properly if it needs auxiliary power.

39 minutes ago, ich777 said:

Sorry but I don't see a Nvidia GPU on your PCIe bus.

 

Are you sure the card isn't defective. Please try to reseat the card and make sure that it plugged in properly if it needs auxiliary power.

 

I wonder if it got fried.  It showed up fine at first, then it ended up getting to around 96c and thats when i first saw the error.  Does / can that happen?

3 minutes ago, mew0717 said:

Does / can that happen?

It can happen but shouldn‘t.

 

Please try the slot that you used before.

still nothing, so i'm getting a new card tomorrow to try again.   Will let you know if that works.

Well new card seems to be working just fine... so problem is solved

Would the nVidia-Driver support the nVidia RTX A4000 ADA SFF graphics card?

11 minutes ago, OrneryTaurus said:

Would the nVidia-Driver support the nVidia RTX A4000 ADA SFF graphics card?

Yes https://www.nvidia.com/Download/driverResults.aspx/224350/en-us/

17 hours ago, SimonF said:

Yes https://www.nvidia.com/Download/driverResults.aspx/224350/en-us/

My wallet hates you but I thank you :)

Went to upgrade to Unraid 6.12.10 today and got these errors, unsure what to do. 

 

Download from plugin package(s): Nvidia Driver for unRAID v6.12.10 failed! Please visit the support thread(s) before rebooting to avoid plugin issues!

 

Nvidia Driver v470.239.06 download failed, please go to the support thread for this plugin and make a post with a screenshot from this error!

I don't know why it tried to download v470.238.06, that seems wrong. I have v550.67 installed currently. The production branch is what I use, and now it says that there is no version found for this version of Unraid. 



Update: tried to just switch to the latest, but it's still trying to download the wrong version. 

Edited May 7, 20242 yr by Indi

25 minutes ago, Indi said:

Download from plugin package(s): Nvidia Driver for unRAID v6.12.10 failed! Please visit the support thread(s) before rebooting to avoid plugin issues!

This is a good indication that something is broken on the Nvidia Driver plugin on your system:

 

Please uninstall the plugin, reboot, install the plugin again, reboot again.

 

After doing that everything should work as it should again.

 

28 minutes ago, Indi said:

and now it says that there is no version found for this version of Unraid. 

On what Unraid version are you, please try the above procedure first before replying.

Just now, ich777 said:

This is a good indication that something is broken on the Nvidia Driver plugin on your system:

 

Please uninstall the plugin, reboot, install the plugin again, reboot again.

 

After doing that everything should work as it should again.

 

On what Unraid version are you, please try the above procedure first before replying.

I am on Unraid 6.12.9 currently, was just moving up to 6.12.10

The below process fixed it up, things seem to be okay now and I am on 6.12.10 

1. Uninstall Plugin
2. Reboot to finish upgrade to 6.12.10
3. Reinstall Plugin
4. Reboot

Hi all, 

I've just updated my system and now my 1070 isn't being used to transcode. I've followed the steps above but smi still shows no processes using the GPU when I am actively playing a show in Jellyfin.

 

Posting here as I feel it's more related to the plugin than Jellyfin, but delete if it's Jellyfins issue. I haven't included logs due to this. Happy to post them if wanted, however.

 

System:
unraid 6.12.10

Nvidia Driver plugin ver 2021.01.19 (using latest driver v550.78)

Jellyfin 10.9.1

GTX1070 8GB

 

I originally updated the plugin and rebooted, then came to the support page and followed the steps below.

 

disabled hardware accel. in jellyfin

uninstalled plugin

rebooted

installed plugin

rebooted

re-enabled hardware accel.

rebooted

played an episode

opened unraid console and checked smi

 

 

Edited May 14, 20242 yr by RancidGnar
Adding comment about lack of logs

25 minutes ago, RancidGnar said:

Posting here as I feel it's more related to the plugin than Jellyfin, but delete if it's Jellyfins issue. I haven't included logs due to this. Happy to post them if wanted, however.

Did you also add this:

 

25 minutes ago, RancidGnar said:

I've followed the steps above but smi still shows no processes using the GPU when I am actively playing a show in Jellyfin.

Are you also sure that you force a transcode, if you just play media and the client is able to play it natively then no transcoding would happen.

To force a transcode, play some media, click on the gear icon in the Jellyfin player and select something really low:

(I would also recommend that you make sure you play some media that is maybe 1080p with a relatively high bitrate so that a transcode is necessary)

 

If you still have issues please also post your settings from Jellyfin itself for transcoding, a full screenshot from your Docker template with Advanced View enabled and your Diagnostics.