[fedora-virt] live migration blocked until dump run in virsh

[Mark McLoughlin]

Hi Gianluca,

On Thu, 2009-08-06 at 15:25 +0200, Gianluca Cecchi wrote:
> With standard F11 x86_64 host I was able to do live migration through
> virsh but not inside virt-manager
> (I got libvirtError: invalid argument in only tcp URIs are supported
> for KVM migrations)
> 
> After enabling fedora-virt-preview repo, I now have:
> 
> gpxe-roms-qemu-0.9.7-4.fc11.noarch
> libvirt-0.7.0-0.9.gite195b43.fc11.x86_64
> libvirt-client-0.7.0-0.9.gite195b43.fc11.x86_64
> libvirt-python-0.7.0-0.9.gite195b43.fc11.x86_64
> perl-Sys-Virt-0.2.0-2.fc11.x86_64
> python-virtinst-0.500.0-1.fc11.noarch
> qemu-common-0.10.91-0.4.rc1.fc11.x86_64
> qemu-img-0.10.91-0.4.rc1.fc11.x86_64
> qemu-kvm-0.10.91-0.4.rc1.fc11.x86_64
> qemu-system-x86-0.10.91-0.4.rc1.fc11.x86_64
> virt-df-1.0.64-2.fc11.x86_64
> virt-manager-0.8.0-1.fc11.noarch
> virt-top-1.0.3-4.fc11.x86_64
> virt-viewer-0.2.0-1.fc11.x86_64
> 
> I'm able to run live migration inside virt-manager: it completes, the
> vm appears running on the
> new host (I can also see the qemu process) but actually it is frozen
> from an OS point of
> view.
> I found a sort of workaround doing this from inside virsh:
> 
>  virsh # list
>  Id Name                 State
> ----------------------------------
>   1 centos53             running
>   2 centos53_node2       running
>   3 prova2               running
> 
> virsh # dump 3 /tmp/dump_prova2.log
> Domain 3 dumped to /tmp/dump_prova2.log
> 
> As soon that the command completes (about 10 seconds with a vm having
> 768MB of ram defined) I have again the vm runni ng and reachable
> through network.....
> Any insight?

That sounds like this bug:

  https://bugzilla.redhat.com/516187

Cheers,
Mark.