Running Unraid on a self-built box, up to date OS, Dockers, apps etc.

I have an RTX A2000 6GB installed, which I use for Jellyfin transcoding and Ollama.

I have to mention that it used to work absolutely fine a few months ago, even with both using the GPU at the same time.

In the last month or so, it just... disappears.

lspci still shows it, but nvidia-smi returns "No devices were found", and the GPU Statistics app shows the RTX A2000 but says "Vendor command returned unparseable data."

At some random interval, it'll re-appear, the GPU Statistics app will show correct data, nvidia-smi will output correctly as well, but as soon as I try to use it within Ollama it'll disappear again.

I've already tried to remove the Nvidia drivers, reboot, disable Docker, re-enable Docker, re-install Nvidia drivers, reboot, disable/re-enable Docker again with the relevant config, and nothing changes.

Short of pulling it out and trying to diagnose the card in another Windows PC, I'm not quite sure what else to try... any ideas, please?

Thank you!

diagnostics-20250820-1858.zip

Edited August 20, 2025Aug 20 by sergiup

I have the exact same problem, albeit with a GTX 1650 instead! It intermittently disappears, and I have also tried all the same steps you've mentioned. This all to say that I am keenly interested in a possible solution!

Edited August 22, 2025Aug 22 by sgraaf

@sgraaf as well - update: I've updated the BIOS (Asrock B550M Pro4 - the A2000 is in the first/main PCIe slot), and for whatever reason, it's now working solidly, for now at least. Honestly, no idea... will keep digging / monitoring and update this thread if this changes.

Edited August 20, 2025Aug 20 by sergiup

Yep I'm having the same exact problem here too RTX 4070Ti. It's quite frustrating.

Update: nope, it's still happening. Worked absolutely fine for a while, now it's acting up again. I have a spare identical A2000 I can try, when I have some time.

I wonder if there is a low power state bug where if the GPU goes idle too long it fails to respond/wake up...

I'd try the oldest driver available and see if the issue persists.

Also Nvidia related discussion should take place in the Nvidia Driver Support topic of the forums.

Edited September 2, 2025Sep 2 by MowMdown

Was there ever a fix to this? Still struggling with this on my 3090ti on 7.2 and 7.1.0rc-1 thank you all!

Same issue, 3090ti. Tracking here or do we need to move this to Nvidia Driver Support topic of the forums?

Edited December 1, 2025Dec 1 by Ed G
Added link to nvidia driver support thread

Oooh, looks like this one might have already been solved on this thread: https://forums.unraid.net/topic/194709-losing-my-gtx1660-gpu-randomly-x-daysweeks-after-a-reboot/#findComment-1589633

tl;dr looks like updating BIOS and disabled power management settings. I am remote and can't access my machine to test, but hopefully it helps someone