RFE: Add functionality to deploy overcloud on vms provisioned using virsh across multiple hypervisors

[smalleni]

Currently, if we want to dpeloy OpenStack at scale using infrared VMs, we are limited by the RAM/capacity on the hypervisor. It would greatly help to add functionality to deploy across multiple hypervisors, that way we can test OSP deployments at scale.

[fhubik]

Hi, as far as I know for these purposes exactly there are hybrid deployments: https://infrared.readthedocs.io/en/latest/hybrid_deployment.html?highlight=hybrid .

[smalleni]

Correct me if I'm wrong but hybrid deployment seem to allow virtual+baremetal deployments but all virtual machines seem to be tied to a single hypervisor.

[saneax]

that is true On Fri, Jun 29, 2018 at 6:59 PM Sai Sindhur Malleni < notifications@github.com> wrote:

Correct me if I'm wrong but hybrid deployment seem to allow virtual+baremetal deployments but all virtual machines seem to be tied to a single hypervisor.

That is true.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/redhat-openstack/infrared/issues/332#issuecomment-401354419, or mute the thread https://github.com/notifications/unsubscribe-auth/AAWgsH81RTXZoqOYYI1fBTTX3Xl721LNks5uBiu_gaJpZM4U8B4y .

-- Sanjay Upadhyay http://saneax.blogspot.com

[obaranov]

Currently the "virsh across multiple hypervisors" feature is not supported by infrared. As @fhubik mentioned you can use for now the hybrid mode. This mode does not allow you to provision several hypervisors, but it allows you to use hypervisors as the baremetal overcloud nodes. So you can have topology like that: 1) Provision minimal virtual topology (e.g. 1 controller) on 1st hypervisior 2) Use the remaining hypervisors as the baremetal overcloud nodes. E.g. you can use another hypervisor as the compute node. And this will give you some performance benefits.