Open Stack

Excerpts from Vitaly Kramskikh's message of 2014-01-14 12:35:43 -0800:
> Hi Jaromir,
> 
> 2014/1/13 Jaromir Coufal <jcoufal at redhat.com>
> 
> > So what we can do at the moment (until there is some way to specify which
> > node to remove) is to inform user, which nodes were removed in the end...
> > at least.
> >
> > In the future, I'd like to enable user to have both ways available - just
> > decrease number and let system to decide which nodes are going to be
> > removed for him (but at least inform in advance which nodes are the chosen
> > ones). Or, let user to choose by himself.
> >
> 
> I think the functionality to choose which node is allocated for each role
> is needed to. I didn't realize how/if it is possible from the wireframes.
> 
> For example, I have 3 racks. I want to allocate 1 node from each rack for
> Controller and make other nodes Computes and then make each rack a separate
> availability zone. In this case I need to specify which exact node will
> become Controller.
> 
> Is it possible to do this?
> 

I think you can do it with rudimentary anti-affinity and affinity rules
in Nova now. However I believe this would be supported by the instance
grouping spec:

https://blueprints.launchpad.net/nova/+spec/instance-group-api-extension

There is also a Tuskar specific blueprint about this that I think would
build on top of the instance grouping.

https://blueprints.launchpad.net/nova/+spec/grouping-baremetal-nodes-for-scheduling

I think that this may have to end up in the Ironic scheduler, as nova
compute nodes will potentially have many baremetal nodes. So a request
will land in Ironic that specifies a desire for a node "not in group
rack_1" or "in group rack_5".