[Pkg-libvirt-maintainers] Bug#539914: Bug#539914: does not know about all instances after libvirtd restart

Guido Günther agx at sigxcpu.org
Tue Aug 4 13:18:30 UTC 2009


severity important

On Tue, Aug 04, 2009 at 02:31:42PM +0200, martin f krafft wrote:
> Package: libvirt-bin
> Version: 0.6.5-3
> Severity: normal
> 
> I am running libvirtd as normal user. If I kill it and restart it,
> it does not know about all the domains it used to control. E.g.
> I only have
> 
>   khyber:~|master|% virsh list                                                                     #1,1528
>   Id Name                 State
>   ----------------------------------
>     3 jugband.madduck.net  running
>     9 lab.madduck.net      running
>   23 mdadm-lab.madduck.net running
> 
> which means lotus.madduck.net and vera.madduck.net are missing (and
> both are running).
> 
> How can I "reconnect" them to libvirtd?
You can't easily (you can shut them down by loging in or via the monitor
sockeet). 
Could you check the logs (~/.libvirt/qemu/log/<vmname>.log) if they
contain anything? Could you also check if
~/.libvirt/qemu/run/<vmname>.xml exists for the missing vms and attach
them to the report?

I assume the VMs are not transient (they do have configuration in
~/.libvirt/qemu/<vmname>.xml)? There was a bug fixed related to daemon
restart and transient VMs.

Is that in any way reproducible? I rechecked with several running VMs and
they all get reconnected back properly.
Cheers,
 -- Guido





More information about the Pkg-libvirt-maintainers mailing list