[Users] Fwd: Re: Using virsh on the node

[Doron Fediuck]

> Subject: Re: [Users] Using virsh on the node
> Date: Mon, 27 Feb 2012 20:25:55 +0200
> From: Itamar Heim <iheim at redhat.com>
> To: Dan Yasny <dyasny at redhat.com>
> CC: users at ovirt.org
> 
> On 02/27/2012 06:34 PM, Dan Yasny wrote:
>>
>>
>> ----- Original Message -----
>>> From: "Nathan Stratton"<nathan at robotics.net>
>>> To: "Dan Yasny"<dyasny at redhat.com>
>>> Cc: users at ovirt.org, "Andrew Cathrow"<acathrow at redhat.com>
>>> Sent: Monday, 27 February, 2012 4:13:54 PM
>>> Subject: Re: [Users] Using virsh on the node
>>>
>>> On Mon, 27 Feb 2012, Dan Yasny wrote:
>>>
>>>> So the engine is a VM in that same cluster, along with the VMs it
>>>> manages?
>>>
>>> Yes, the only thing I have not been able to figure out is how to let
>>> the
>>> ovirt-engine know that ovirt-engine is running. :) I thought that it
>>> would
>>> see via libvirt, but it does not. I also tried setting it in the db,
>>> but
>>> it keeps getting cleared. Any idea on how to let ovirt-engine know
>>> that a
>>> host has been started manually?
>>
>> Seriously, not a good idea. The Engine is supposed to be looking at the setup, and not be in the setup, where it may be affected by the problems that  might occur to the setup.
>> I'm pretty sure there's lots of though being devoted to removing this limitation, but when we say it's a bad idea, it really is, it's not like we simply want you to run an extra machine
> 
> Doron - this reminds me of the all-in-one discussion. what was the
> resolution on flagging the node as a node running an engine VM (even
> without engine knowing about its own VM)
> 
Sorry for breaking the thread, technical issues.
Currently we handle standard platforms (ie- not the node),
by using the all-in-one rpm which is currently in work.
So when you add a host via the REST you may specify not
to reboot it, and you should not specify power-management
from properties, which will make sure fencing will not occur.

Other than that there's no specific awareness to the host
being an angine-core host.

>>
>>>
>>>> What do you use for storage?
>>>
>>
>> That's easier than block storage, but still, what happens if the host where the engine was running failed - who would update VM statuses and restart the HA VMs?
>>
>>> NFS:
>>>
>>> mount 10.13.0.4:/images/ /rhev/data-center/mnt/10.13.0.4:_images
>>> mount 10.13.0.4:/iso/ /rhev/data-center/mnt/10.13.0.4:_iso
>>>
>>>> Dan Yasny
>>>> Red Hat Israel
>>>> +972 9769 2280
>>>
>>>> <>
>>> Nathan Stratton                                CTO, BlinkMind, Inc.
>>> nathan at robotics.net                         nathan at
>>> blinkmind.com
>>> http://www.robotics.net
>>>                         http://www.blinkmind.com
>>>


-- 

/d

"All computers wait at the same speed."