Hmm, maybe because now the driver override for vfio-pci works? Inside the VM, it’s the nvme driver of course.
I edited my post btw. While what I had works, the saner option is to use lspci -D to see the slot including domain, instead of assuming 0000: for the PCI domain. On single-CPU boards it will always be 0000:, but not every board is single CPU. And I had misnamed the thing as the “id”, it is of course the slot. The “id” is what lspci -nn shows as [<vendor>:<device>]
Just a litle update on this.
I kept getting crashes that would take the whole VM down and messed up TN a bit too with extremly slow shutdowns. I did a bit more research on the drive I was using. Seems WD drives in general seem to have issues being passed through to VM’s. A few people reporting similar issues on Proxmox.
So I have swapped it out with a P41 Platinum I had kicking about. Only done that start of this week and no crashing so far.
Anyone coming to this thread in the future, watch out if you use a WD NVME.
