You are not logged in.
Since the latest update to qemu-kvm 1.1.1-1 my Linux VMs do not load and qemu-kvm shows 100% CPU usage until killed (I waited long).
The VMs in problem are: Arch 32-bit and CentOS 5.8 64-bit. Both are up to date.
Both VMs reside in the file images (QCOW2).
CentOS uses disk virtio, Arch uses only net virtio.
The issue can be solved running without the -nographic option. I do not like that and prefer connecting to VMs via SSH.
The issue can be solved also running with the -no-kvm option, but this is not an option :).
I assume that the VMs fail somewhere early, so that I even do not see FS journal recovery messages on the next load.
Downgrading qemu-kvm to 1.1.0-1 didn't solve the situation, but solved downgrading to 1.0.1-3.
I think I missed something with the new 1.1 branch. Does anybody have and idea how can this be fixed?
P.S. The options are trivial:
-boot c -k en-us -daemonize -nographic -enable-kvm -m 256 -drive file=arch.img,index=0,media=disk -net nic,macaddr=...,model=virtio -net tap,ifname=tap1
Last edited by Synchrotron (2012-08-16 23:41:42)
Offline
Found a lengthy discussion of this issue here - https://bugs.launchpad.net/qemu/+bug/1021649
-no-kvm-irqchip solved the problem for me.
-global virtio-blk-pci.ioeventfd=off did not.
Offline