Error in one of my servers

[Armando Ramos Roche]

Hi all
I have 3 servers HP Proliant G8, all have Promox 8.2.4 in a cluster..
One of them recently show this error in syslog:

Sep 13 14:47:05 pve2 kernel: kvm_intel: kvm [1295]: vcpu7, guest rIP: 0xfffff8034f7e7fa3 Unhandled WRMSR(0x1d9) = 0x1
Sep 13 14:47:05 pve2 kernel: kvm_intel: kvm [1295]: vcpu7, guest rIP: 0xfffff8034f7e7fa3 Unhandled WRMSR(0x1d9) = 0x1
Sep 13 14:47:05 pve2 kernel: kvm_intel: kvm [1295]: vcpu7, guest rIP: 0xfffff8034f7e7fa3 Unhandled WRMSR(0x1d9) = 0x1
Sep 13 14:47:05 pve2 kernel: kvm_intel: kvm [1295]: vcpu7, guest rIP: 0xfffff8034f7e7fa3 Unhandled WRMSR(0x1d9) = 0x1
Sep 13 14:47:05 pve2 kernel: kvm_intel: kvm [1295]: vcpu7, guest rIP: 0xfffff8034f7e7fa3 Unhandled WRMSR(0x1d9) = 0x1
Sep 13 14:47:05 pve2 kernel: kvm_intel: kvm [1295]: vcpu7, guest rIP: 0xfffff8034f7e7fa3 Unhandled WRMSR(0x1d9) = 0x1

Any can helpme pls

 

[dakralex]

If nothing else happened (e.g. the VM failed to start/crashed), these are harmless warning messages that are caused when a QEMU guest is trying to write to a register that might be present in the host CPU, but is not exposed to the QEMU guest itself. This can happen for intentional unhandled WRMSRs (write to model-specific register), e.g. the QEMU guest tries to disable a CPU feature on the physical CPU which should obviously not be possible.

In your case, I'm guessing that a program in one of your VMs is trying to write to the IA32_DEBUGCTL (for Intel CPUs) or DebugCtl (for AMD CPUs), which has the MSR address 0x01d9, which you can source from the respective vendor's architecture manuals. If there was no intention for this, then you can safely ignore these message and also diable them with the following command as specified at [1]:

echo "options kvm ignore_msrs=1 report_ignored_msrs=0" > /etc/modprobe.d/kvm.conf

[1] https://pve.proxmox.com/wiki/PCI_Passthrough#Tips