You could try disabling IOMMU in the board BIOS to see it makes any difference, assuming it's not needed.

Okay, so my optimism was a bit premature ;-)
Today I decided to spin up my VM (the one called Jabberwocky that has a GPU passed through) and all hell broke loose. First off, one of the SSD's (sdi) from my ZFS pool named 'pppool' dropped offline. I have attached the info from the pool status as it was displayed.

Besides that, three hard drives from my array dropped away and were listed under Unassigned devices. In panic, I shut down the VM and then the server completely and after booting back up, luckily everything was back to normal again. Note that I no longer have that VFIO bind error that I was getting before, but apparently still something is not right with my configuration. I now have it configured as follows:

Note that these two devices are the only devices being passed through to the VM and are also the only two devices in the device list with the 10:00 device identifier (or however that should be called).
One other thing that might be important is that under the VM config the following is stated: GPU is primary adapter, vbios may be required.
I am not sure how to check which video adapter (also have an ARC A310 in the system) is the primary one. The VM itself did spin up without any issues.

I am also attaching the diagnostics file. Really hoping someone can help me out and point me in the right direction to find out what is wrong with my system.
Thanks for any thoughts!

pppool status info.txt olifant-diagnostics-20250713-1502.zip

On 6/30/2025 at 12:48 PM, JorgeB said:

You could try disabling IOMMU in the board BIOS to see it makes any difference, assuming it's not needed.

Hi Jorge, somehow I missed this reply. Is IOMMU not needed for hardware passthrough?

16 hours ago, Sten3danny said:

Hi Jorge, somehow I missed this reply. Is IOMMU not needed for hardware passthrough?

It is.

Jul 13 14:44:04 Olifant kernel: ahci 0000:0e:00.0: AHCI vers 0001.0301, 32 command slots, 6 Gbps, SATA mode

Jul 13 14:44:04 Olifant kernel: ahci 0000:0e:00.0: 4/4 ports implemented (port mask 0x33)

Jul 13 14:44:04 Olifant kernel: ahci 0000:0e:00.0: flags: 64bit ncq sntf ilck pm led clo only pmp fbs pio slum part

Looks like an issue with this controller; it got detected again, like if it had dropped, though I can't see when.

Thx Jorge,

that would make sense because the drives that dropped away are exactly the drives connected to that controller:

I need to double check, but I am pretty sure this is an onboard controller on the motherboard. The issue basically occured the minute I started up that VM, I will check later if I can reproduce the issue.
Could it be related to the fact that the graphics card I am passing through is the primary card (according to Unraid), and would uploading a vbios file help?

Honeslty I've never really fully understood how this passthrough works and what I should pay attention to. It has mostly been a trial and error process, which has worked so far, but now it seems I'm finally at the point where I need to put some serious effort into it :-)

1 hour ago, Sten3danny said:

The issue basically occured the minute I started up that VM

That makes me think the VM may be trying to passthrough that device, which VM since there are multiple ones?

1 hour ago, JorgeB said:

That makes me think the VM may be trying to passthrough that device, which VM since there are multiple ones?

It is Jabberwocky.
Indeed you are right, that could be, but I cannot understand why. I only want to pass through the graphics card and sound card (10:00.0 and 10:00.1), which I have selected in the VM config.

Yep, looks good to me. Do you have ACS override enabled? That can sometimes cause strange issues.

26 minutes ago, JorgeB said:

Yep, looks good to me. Do you have ACS override enabled? That can sometimes cause strange issues.

No, it is disabled.
Could the graphics card being the primary adapter (according to Unraid) be causing issues?
I don't know how to change that, other than maybe by physically swapping my two graphics cards, which I might try if it could potentially help..

36 minutes ago, Sten3danny said:

Could the graphics card being the primary adapter (according to Unraid) be causing issues?

It shouldn't, but try to start the VM again and see if it causes the same issue, to make sure it's related to that.

1 hour ago, JorgeB said:

It shouldn't, but try to start the VM again and see if it causes the same issue, to make sure it's related to that.

I will do that, once the server has a quiet moment.

Thx again Jorge!

Finally got around to testing this, so I spun up my VM again and guess what, nothing happened! 😁
I actually shut down and restarted the VM a few times and everything has been stable so far.

I am of course not complaining..

Guess I will just enjoy this unexpected succes for now and wait until the next 'incident' ;)