Processor Utiiization Sky Rockets -- Hard reset needed

rayk_sland

Active Member
Jul 30, 2009
53
1
28
This was all referenced in a previous post, and I did my best to follow all the suggestions and the problem seemed to go away. But it came back last night with a vengeance.

Latest stable proxmox 1.6 kernel
64 bit Debian Lenny guest.
Virtio Disk on a LVM partition from a shared pool (exactly as recommended)
3 processor cores

Twice in a row I had to issue a qm stop command against it. No shut down possible. Twice in a row the guest had to check the damaged file system and reboot a second time.

qm monitor 'info cpus' command showed 2 of 3 CPU's in an ?inactive? state (sorry I should have documented exactly what it said -- I was in a hurry to get it back up, at any rate 3 CPU's were listed and two had an extra flag something to do with inactive) Anyways I reconfigured the guest for only one core and it seems to be stable now. Any Idea what causes this behaviour?
 
name: MailServer-64
ide2: local:iso/debian-502a-amd64-netinst.iso,media=cdrom
sockets: 1
vlan0: e1000=26:D2:30:41:CA:CD
bootdisk: virtio0
# virtio1: OCFS_Share:108/vm-108-disk.qcow2
ostype: l26
memory: 9000
vlan1: e1000=1E:1E:E1:C9:75:04
onboot: 0
cores: 1
boot: cd
freeze: 0
cpuunits: 1000
acpi: 1
kvm: 1
virtio0: pool2:vm-108-disk-1
args: -no-hpet


after reading the other similar thread,(http://forum.proxmox.com/threads/5144-100-CPU-on-host-VM-hang-every-night) I added the args: -no-hpet. line. (and the kernel parameter in the guest grub.conf) -- they're is not active yet as I haven't had a good opportunity to restart the server (it's pretty critical)
the cores: line used to read 'cores: 3' but reducing it to 1 has given it the stability I referred to.