Fangtooth VM 100% CPU usage

TrueNAS General
, , ,
1

New VM on Fangtooth running Windows Server 2025. Everything inside the VM feels really sluggish. I just noticed that the VM uses 100% of a single core, even if it’s not doing anything.

From running top:

    PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
   9624 nobody    20   0   17.3g  16.0g  15.9g S  97.4   6.4  77:48.50 qemu-system-x86

Now, there’s one possibility that I’m shooting myself in the foot by doing nested virtualization. I’m in the process of moving away from Proxmox (TrueNAS as a VM) to just hosting everything on bare metal TrueNAS. Nothing here is mission critical, as it’s just a homelab for self-hosting. However, I really, really need a Windows VM, so I need to test if I can get it working inside TrueNAS before I get rid of the one hosted by Proxmox. Is it possible that the CPU usage is because of this, and I can safely ignore it?

CPU: EPYC 7C13
MB: Supermicro H12SSL-i

2

What do you see inside the Windows VM? Is there a process using 100% of a virtual CPU? How many virtual CPU’s did you give to this VM?

Wait …

So you are running TN as a VM and trying to run a VM inside TN?

I will bet no one has tested that configuration.

3

I know the person who could.

1 Like
4

Well, you pointed me in the right direction, making me double-check the CPU usage inside the VM. It’s as low as I remembered it, but it seems I messed up the CPU allocation, accidentally giving it “a few” more cores than I intended. Sure enough, there’s one core’s worth of interrupts.

I guess this is a completely different issue, so this topic can be closed/removed. Will create a new one about the interrupts if I can’t figure it out.