Open Stack

On 04/24/2013 09:51 AM, Leen Besselink wrote:
> On Wed, Apr 24, 2013 at 09:35:19AM -0400, Russell Bryant wrote:
>> On 04/23/2013 08:20 PM, Robert Collins wrote:
>>> +1 on using Heat to orchestrate the response to a node failure.
>>
>> Yeah, it's quite clever, I like it.  It's as if you TripleO folks are on
>> to something here.  ;-)
>>
> 
> I thought the same thing, until I realized that maybe not everyone wants to setup
> OpenStack on OpenStack. But might still want VM HA.

If folks want to diverge from "the OpenStack way", that's perfectly
fine.  In that case we just need to make sure all of the primitives are
there for someone to implement their own solution.

As far as I know, the compute API already exposes the primitives
necessary here.  If you detect a compute node failure using your Thing,
you can use live-migrate or evacuate, depending on whether the compute
node is on its way to dying, or is already dead.

-- 
Russell Bryant