[one-users] libvirt persisent vs transient domains

Ruben S. Montero rubensm at dacya.ucm.es
Fri Feb 18 09:28:55 PST 2011


OpenNebula monitors VMs in unknown state. So, if the VMs are restarted,
OpenNebula should be able to recover those VMs and move them from unknown to
running. The requirement is to and the id's are preserved at the hypervisor
level (e.g. autostart i think)



On Fri, Feb 18, 2011 at 6:05 PM, Shi Jin <jinzishuai at gmail.com> wrote:

> Thank you for the discussion.
> Yes, there is definitely a possible problem of out of sync between the host
> and OpenNebula.
> For example, after the node power failure, the one  status for the VMs
> would be unknown. If people then deleted the VM, restarting the VM upon
> power resumption would cause a problem.
> But this seems to be a more operational problem: just don't delete the VMs
> on a power failure node.
> There are many other ways to cause out of sync under existing OpenNebula
> setup and it is up to the administrator's operation to ensure everything is
> consistent.
> Also, starting a power failure VM definitely has the potential problem of
> disk corruption. But this is the best we can do upon such a tragic
> situation. One should ask more on why on the earth did the power fail in the
> first place? The autostart setup would simply ease a little bit of the pain.
> Shi
> On Thu, Feb 17, 2011 at 8:42 PM, Manikanta Kattamuri <
> mani.kattamuri at hexagrid.com> wrote:
>> Assume all this is set and the vm's automatically come up after the
>> servers get online.
>> would it not create more problems then solve as you one is not in sync
>> with all the vm states?.
>> If there are any monitoring tools configured based on one status, will
>> report failure.
>> Starting of the vms through oned will not succeed or cause disk
>> corruption?
>> Mani.
>> On Fri, Feb 18, 2011 at 4:03 AM, Shi Jin <jinzishuai at gmail.com> wrote:
>>> Hi there,
>>> Right now, all KVM (I think Xen too but I don't know it for sure) VMs are
>>> started as a transient libvirt domain using the "virsh create <xml>"
>>> command.
>>> There is nothing wrong with it but recently I had a facility wide power
>>> failure and all VMs running on the node are gone upon the server shutdown. I
>>> had to manually start all the domains that were running at the point of
>>> failure after power is resumed. Then I thought that if I put the VMs as
>>> libvirt autostart domains, they will be automatically started upon boot,
>>> which would save me a  lot of manual process. But to do that, I have to make
>>> the VMs persistent libvirt domains using the "virsh define" and "virsh
>>> start" commands in sequence instead of the single "virsh create".
>>> Is this a good idea? Comments welcome.
>>> Thanks.
>>> Shi
>>> --
>>> Shi Jin, Ph.D.
>>> _______________________________________________
>>> Users mailing list
>>> Users at lists.opennebula.org
>>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
> --
> Shi Jin, Ph.D.
> _______________________________________________
> Users mailing list
> Users at lists.opennebula.org
> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org

Dr. Ruben Santiago Montero
Associate Professor (Profesor Titular), Complutense University of Madrid

URL: http://dsa-research.org/doku.php?id=people:ruben
Weblog: http://blog.dsa-research.org/?author=7
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20110218/af25ff17/attachment-0003.htm>

More information about the Users mailing list