[one-users] Instance migration based on resources
Ruben S. Montero
rsmontero at opennebula.org
Wed Jan 16 09:32:11 PST 2013
Hi
We've considered this a couple of times in the past, and first, the support
from the hypervisor and (most notably) guest OS'es is not uniform. And
second, we have the work needed to make this happen. As *very accurately*
:) described by Simon this implies recalculating quotas, implications on
the current accounting schemes...
This is for sure a nice feature, and we'll keep it in our radar for future
releases.
Meanwhile, I think that including a workaround for offline re-sizing based
on persistent images is a great idea. We'll include that in the
documentation. Thanks Simon for the suggestion.
Best
Ruben
On Wed, Jan 16, 2013 at 4:59 PM, Simon Boulet <simon at nostalgeek.com> wrote:
> It's definitively possible to do offline upgrade/downgrade with
> OpenNebula. You need to cancel the VM and submit a new template reusing the
> image. The main problem is the image, you need to make it persistent.
>
> Making hot upgrade/downgrade of CPU and memory is not currently possible
> directly in OpenNebula (you have to cheat by doing the changes manually on
> the hypervisor, most hypervisors support hot plugging CPU and memory).
> Adding that feature in OpenNebula would require quite a bit of work,
> notably with the quota feature, perhaps to the scheduler too (migrate if
> you try to hotplug more CPU or memory than the host has available), and
> also for people using the OpenNebula template description to
> do resource based billing. Currently the template is static and never
> change. You simply have to parse the template the first time and then look
> at the VM history records to properly bill your user. Making the template
> attributes change would probably require some resources history records
> (knowing at what time the VM resources changed and bill the user
> accordingly).
>
> I'm not saying it's not something possible, only that it may be more work
> that a lot of us think.
>
> If you have your image set up properly, you can do offline
> upgrade/downgrade with OpenNebula with minimal downtime for the VM.
>
> Perhaps a page on the OpenNebula documentation could be created to explain
> the situation and show how to do offline upgrade/downgrade with OpenNebula.
>
> Simon
>
>
>
> On Wed, Jan 16, 2013 at 10:35 AM, christopher barry <cbarry at rjmetrics.com>wrote:
>
>> On Tue, 2013-01-15 at 23:27 +0000, André Monteiro wrote:
>> > Hi Gandalf,
>> >
>> >
>> > If you don't want to destroy the VM for some reason, check my message
>> > on
>> >
>> http://lists.opennebula.org/pipermail/users-opennebula.org/2012-August/020021.html
>> >
>> >
>> > --
>> > André Monteiro
>> >
>>
>> reading that thread it appears it was evaluated to potentially be a
>> feature (cold update). Did it not get implemented yet? Ideally, hot
>> update would be the thing to have. Isn't hotplug cpu and memory
>> supported in kvm now?
>>
>> -C
>>
>>
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opennebula.org
>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>>
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opennebula.org
> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>
> --
> Ruben S. Montero, PhD
> Project co-Lead and Chief Architect
> OpenNebula - The Open Source Solution for Data Center Virtualization
> <http://lists.opennebula.org/listinfo.cgi/users-opennebula.org>
> www.OpenNebula.org | rsmontero at opennebula.org | @OpenNebula
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20130116/7ea3eafd/attachment-0002.htm>
More information about the Users
mailing list