CPU usage on VM is reported higher when logged into Proxmox UI

chrcoluk

Renowned Member
Oct 7, 2018
308
89
93
47
I have noticed fairly consistently when logged into the Proxmox GUI, that CPU usage graphs on a per VM basis report higher utilisation, this makes no sense. Only the host should have the higher utilisation.

Not sure why I am posting this, but I guess I am looking into insight as to why it happens.
 
Does the reported CPU usage increase occur across VMs running different guest operating systems, such as Windows and Linux?
 
Check whether the extra CPU is used inside the guest or only by the QEMU process on the host. Run `top` inside the VM and `top -p $(cat /var/run/qemu-server/<vmid>.pid)` on the node, first with no browser tab open, then with the VM's Summary page open, then with its Console open.

If `qemu-ga` rises inside the guest only while the Summary page is open, it's likely the guest agent calls the page makes for the IP addresses and OS details.

If only the QEMU process on the host rises while the Console is open, it's likely the display emulation redrawing for the noVNC session.

Drafted with AI.
 
Yes the OS running doesnt matter, all that matters is if the I am logged into proxmox. No guest installed, I will check the QEMU process itself.
 
Last edited:
Yeah it doesnt need a open web console. Also I seem to only need to be logged in to proxmox, I dont need to be on the VM page. I havent done your previously requested test, I apologise as I got busy for a while.
 
Both things can be true, but for what you're describing — CPU jumps just from being logged into the UI, console closed — it's almost certainly just the web interface polling overhead.

When you're signed into Proxmox, the backend keeps hitting every VM's QEMU process via QMP to pull status, metrics, config info, all that. All that CPU time from the qemu process gets lumped into the VM's usage graph. So even a totally idle guest will show higher numbers as long as the UI is open somewhere. It's real overhead from monitoring, not actual load inside the VM, and it drops right back when you log out.

That said, there's a second reason the numbers never line up perfectly: the Proxmox figure also includes time the vCPU threads spend waiting on the host scheduler. If the node itself is busy, an idle guest can look busier than it is. And with Windows guests? Forget about it — Task Manager counts completely differently, they'll never match.

Easy way to confirm: log all the way out of the UI, close the tabs, then watch the qemu-system process for that VM in top on the host. If it drops immediately, it's just UI polling — nothing broken. If it stays high, then dig into the vCPU threads and check inside the guest.
 
  • Like
Reactions: Johannes S