Hand holding needed - too confused to solve a problem!

TrueNAS General
, , ,
1

Hello

For use with an official app, I tried a new Nvidia graphics card in place of an existing Nvidia card but it seems that various UUIDs became muddled. I wrote some detail regarding the errors I saw here

I also spotted something - a known trueNAS bug - which seemed similar but I am unable to decide whether this recognised bug might be the cure for my Nvidia UUID confusion.

Please can someone let me know if (best guess) the two are related such that it’s worth me attempting the fix discussed? I really don’t want to risk messing up my app by being unable to revert to the current state.

2

It sure looks similar to the known bug issue, but you can see, that ticket is a Low Priority.

You can try this for now…

  1. File a bug report, make sure you include the /var/log/nvidia-installer.log as well as all other bug report files, and the hardware details asked for in Honeybadger’s posting. I’m not sure that I would classify it to be the exact same problem, however it looks related. Hopefully this will increase the priority.
  2. Update to the current EEL version, you are behind by 1.
  3. If you have not upgraded the ZFS Feature Set, you can roll back to the previous version and wait for something to happen, like 25.04 coming out.

With that said, I have said this many times before to people, Never Upgrade your ZFS Feature Set unless you need something that is included in the feature set. 99.9% of the time, for a home user the upgrade contains nothing of value to them. Do not let the alert throw you off.

1 Like
3

Thanks Joe. I’ll submit my bug report too (and I have now updated TrueNAS) and I think I shall take the plunge with the “repair” that’s mooted by @HoneyBadger seeing as you also think it looks similar. It’s worth a go I reckon.

update: I fitted the new GPU and then I ran the command and got the relevant strings:

root@truenas[~]# midclt call app.gpu_choices | jq 
{
  "0000:01:00.1": {
    "vendor": null,
    "description": "Matrox Electronics Systems Ltd. MGA G200EH",
    "vendor_specific_config": {},
    "pci_slot": "0000:01:00.1"
  },
  "0000:07:00.0": {
    "vendor": "NVIDIA",
    "description": "Quadro P620",
    "vendor_specific_config": {
      "uuid": "GPU-9bd360c8-7be6-3874-6ff1-96b3d0f122c9"
    },
    "pci_slot": "0000:07:00.0"
  }
}

which enabled me to do

root@truenas[~]# midclt call -job app.update frigate '{"values": {"resources": {"gpus": {"use_all_gpus": false, "nvidia_gpu_selection": {"0000:07:00.0": {"use_gpu": true, "uuid": "GPU-9bd360c8-7be6-3874-6ff1-96b3d0f122c9"}}}}}}'

which resulted in

image
image1982×277 39.1 KB

and, after several app restarts during which trepidation and anxiety hit very high peak values, I got Frigate going with the new GPU.