[SOLVED] ZSF Snapshot: unable to parse value of 'runningcpu' - got undefined value

[Foxi352]

Hi,

I tried everything but i can't get SnapShots of my VM 100 to work. It is on ZFS.
The error is "unable to parse value of 'runningcpu' - got undefined value".

I need a hint that points me to the direction to continue digging ... I am out of ideas.

A second problem is that every failed snapshot try leaves a vm-<id>-state-<snapshot_name> file of ~ 50GB

Here my pveversion output and my machine config file.

pve-manager/6.2-10/a20769ed (running kernel: 5.4.44-2-pve)

args: -enable-kvm -device isa-applesmc,osk="XXX" -smbios type=2 -cpu Penryn,kvm=on,vendor=GenuineIntel,+vmx,+kvm_pv_unhalt,+kvm_pv_eoi,+invtsc,vmware-cpuid-freq=on,+pcid,+ssse3,+sse4.2,+popcnt,+avx,+aes,+xsave,+xsaveopt,check -device usb-kbd,bus=ehci.0,port=2
balloon: 0
bios: ovmf
boot: c
bootdisk: sata0
cores: 8
cpu: host,flags=+pdpe1gb;+hv-evmcs;+aes
hostpci1: 09:00,pcie=1
hostpci2: 04:00,pcie=1,x-vga=1
hostpci3: 00:14.0,pcie=1,rombar=0
hugepages: 1024
machine: pc-q35-4.0.1
memory: 24576
name: MacOS
net0: vmxnet3=CE:20:D5:F3:BA:C2,bridge=vmbr1,tag=100
numa: 1
onboot: 1
ostype: other
sata0: MacOS:vm-100-disk-0,cache=unsafe,size=850G,ssd=1
scsihw: virtio-scsi-pci
smbios1: uuid=XXXXXXX
sockets: 1
tablet: 0
vga: none
vmgenid: XXXXXX

 

[fiona]

Hi,
I can reproduce this here. I'll see if I can find the cause. Thanks for reporting!

EDIT: patch is here

 

[Foxi352]

Hi,
I can reproduce this here. I'll see if I can find the cause. Thanks for reporting!

EDIT: patch is here

Great. Thank you for your quick response. I applied the patch but still have the same error message. Do i have to restart the host for this to work ?

How can i get rid of all these 50GB -state- files from the failed snapshot attempts ? If i want to delete the files i get Cannot remove image, a guest with VMID '100' exists! You can delete the image from the guest's hardware pane

 

[fiona]

Reboot isn't necessary, it should be enough to restart the PVE daemon with systemctl restart pvedaemon.service.
To remove the state files, use pvesm free <storage-ID>:vm-100-state-<name of failed attempt> or remove them manually.

 

[Foxi352]

Both worked as described !

Vielen Dank !

 

[admarty]

Hi,
I can reproduce this here. I'll see if I can find the cause. Thanks for reporting!

EDIT: patch is here

Hello.
I faced the same problem when adding args -cpu host,enforce,+kvm_pv_eoi,+kvm_pv_unhalt,model-id='Intel(R) Xeon(R) CPU D-1537 @ 2.10GHz',-vmx
I tried editing the CPUConfig.pm file as follows but it still didn't fix the problem.
our $qemu_cmdline_cpu_re = qr/^((?>[+-]?[\w\-\._='@]+,?)+)$/;

Here my pveversion output and my machine config file.
pve-manager/8.0.4/d258a813cfa6b390 (running kernel: 6.2.16-15-pve)

acpi: 1
agent: 1
args: -cpu max,enforce,+kvm_pv_eoi,+kvm_pv_unhalt,model-id='Intel(R) Xeon(R) CPU D-1537 @ 2.10GHz',-vmx -smbios type=1,product='XG-1537 1U',manufacturer=Netgate
balloon: 128
boot: order=scsi0
cores: 2
hotplug: network,usb
machine: q35
memory: 256
meta: creation-qemu=7.1.0,ctime=1673178426
name: OpenWrt
net0: virtio=76:F6:38:90:83:06,bridge=vmbr1,firewall=1,tag=110
net1: virtio=4E:62:58:FA:FD:74,bridge=vmbr0,firewall=1
numa: 0
onboot: 1
ostype: l26
scsi0: NVME-thin:vm-110-disk-0,discard=on,iothread=1,size=124M
scsihw: virtio-scsi-single
serial0: socket
smbios1: uuid=2f939666-485f-43e4-a0f4-38b02ac139bd
sockets: 1
tablet: 0
tags:
vga: serial0
vmgenid: 187940c9-494a-486b-a5d9-43358f7f034c

 

[fiona]

Hi,

Hello.
I faced the same problem when adding args -cpu host,enforce,+kvm_pv_eoi,+kvm_pv_unhalt,model-id='Intel(R) Xeon(R) CPU D-1537 @ 2.10GHz',-vmx

custom args are just provided as an escape hatch, there's no guarantee Proxmox VE will handle everything correctly.

I tried editing the CPUConfig.pm file as follows but it still didn't fix the problem.
our $qemu_cmdline_cpu_re = qr/^((?>[+-]?[\w\-\._='@]+,?)+)$/;

I think your still missing the parentheses ( ) from Intel(R) Xeon(R) as allowed characters.

But feel free to open up a feature request on our bug tracker and please describe your use case there (i.e. why you need the explicit model-id): https://bugzilla.proxmox.com/

 

[admarty]

Hi,

custom args are just provided as an escape hatch, there's no guarantee Proxmox VE will handle everything correctly.


I think your still missing the parentheses ( ) from Intel(R) Xeon(R) as allowed characters.

But feel free to open up a feature request on our bug tracker and please describe your use case there (i.e. why you need the explicit model-id): https://bugzilla.proxmox.com/

Great news! I managed to resolve the issue. I updated the $qemu_cmdline_cpu_re regex pattern to include parentheses and additional allowed characters, like so:
our $qemu_cmdline_cpu_re = qr/^((?>[+-]?[a-zA-Z0-9_ \-\._=@()]+,?)+)$/;

This change successfully fixed the problem I was facing when adding the custom args. Thanks for your earlier suggestion, and now everything is working as expected.