KVM constant CPU hog (Win 2008 R2 x64 guest)

[gkovacs]

One of our PVE servers hosts several OpenVZ and one freshly installed KVM VM, with a Windows Server 2008 R2 x64 guest (with latest VirtIO for net and disk).
Hardware is Q6600, 8GB RAM, mdraid on Intel SATA.

If more than one CPU core is enabled for the KVM VM, then it constantly eats 30% CPU even when idle.
When only one CPU core is enabled, it idles around 1-5%.

Is this related to DEP (Data Execution Prevention) or x64?

proxmox2:~# pveversion -v
pve-manager: 1.9-26 (pve-manager/1.9/6567)
running kernel: 2.6.32-7-pve
proxmox-ve-2.6.32: 1.9-55+ovzfix-2
pve-kernel-2.6.32-4-pve: 2.6.32-33
pve-kernel-2.6.32-6-pve: 2.6.32-55+ovzfix-1
pve-kernel-2.6.32-7-pve: 2.6.32-55+ovzfix-2
qemu-server: 1.1-32
pve-firmware: 1.0-15
libpve-storage-perl: 1.0-19
vncterm: 0.9-2
vzctl: 3.0.29-3pve1
vzdump: 1.2-16
vzprocps: 2.0.11-2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.15.0-2
ksm-control-daemon: 1.0-6

Is this going to get fixed? I remember that SMP has always been problematic with KVM, but thought this has been resolved.

Anyone else having this same problem?

 

[daveo]

I have noticed this too. Client needed to have a Quad core solution so I enabled both 4 sockets 1 core and 1 socket 4 cores. In both situations the host show high CPU utilisation that is not backed by activity in the guest.

I would be curious to know if anyone has found a work around?

 

[artisticab]

I had this problem too, it made my 2003 x86 Server guests cpu usage max out. I am still learning linux and proxmox, but I figured out if I changed the running kernel back to "2.6.32-6-pve" instead of "2.6.32-7-pve" the problem goes away. So that is what I did in my situation. I used webmin to change the boot order of the kernels.

I don't think there will be a fix, as proxmox 2.0 is already out and everyone is migrating to the newest version. Hope this helps.