a673e65006
- rsd-virt-for-nova Change-Id: Ib6653189d98822f9dfd5e635df8fe3e6d4db19d6 Signed-off-by: Helena McGough <helena.mcgough@intel.com>
4.0 KiB
User guide
Introduction
This project is a wip. It is a new nova-virt driver that allows the management an IntelRSD deployment software architecture. This architecture deployment facilitates the orchestration of a composable infrastructure through OpenStack. The new driver itself enables the management of RSD composable nodes through the use of nova compute service. Therefore you can manage the deployement of a composed node like you would a VM, container or a baremetal instance, etc.
To set up your OpenStack deployment to use the new virt driver follow the instructions provided in installation_guide.rst.
Usage Instructions
The new rsd-virt-for-nova virt driver aims to use the same CLI commands as any of the other standard virt drivers.
Currently the following commands are supported by the new virt driver:
openstack server create --image <IMAGE_ID> --flavor <FLAVOR_ID> <INSTANCE_NAME> openstack server delete <INSTANCE_NAME> openstack server start <INSTANCE_NAME> openstack server stop <INSTANCE_NAME> openstack server reboot --hard <INSTANCE_NAME> openstack server reboot --soft <INSTANCE_NAME> openstack server reboot <INSTANCE_NAME>
To create an instance of type composed node through OpenStack you have to use one of the specific RSD flavors that are automatically generated based on the resources available in the RSD deployment. These flavors can be identified by their name and the extra_specs used to define them when created. The extra_specs define the custom resources used by the resource provider to track resource consumption through the placement API.
Field Value OS-FLV-DISABLED:disabled OS-FLV-EXT-DATA:ephemeral access_project_ids disk id name os-flavor-access:is_public properties ram rxtx_factor swap vcpus False 0 None 0 785919MB-32vcpus RSD-785919MB-32vcpus True resources:CUSTOM_1_S_3='1' 785919 1.0
32
In general a flavor can only be used once depending on the amount of composable systems available in the RSD deployment. They will define the specific system that will be consumed from the RSD deployment.
Usage Tracking
The consumption of resource can be tracked through nova hypervisors
used to boot the composed node instances from. These are defined at the
Chassis level of the RSD deployment. It can also be tracked
in the placement API in terms of resource providers and their inventory.
There is a resource provider for each of the hypervisors at the
Chassis level and the a child resource provider for each
System contained within the defined
Chassis.