This session will include the following subject(s):
OpenStack in OpenStack:
In this presentation, we discuss how to deploy a multi-node OpenStack environment inside a typical openstack environment, where you usually get a VM with an internal IP and optionally a public floating IP via NAT. Such a virtual deployment option is desirable for developing and testing new features that require multiple node setup, such as networking, high availability, live upgrade, etc.
We demonstrate a few key enabling technologies:
* nested server virtualization; * software-defined networking and policy routing for virtual private and public layer-2 networks; * chef for automatic configuration and deployment.
(Session proposed by Yun Mao)
multi-node openstack testing:
Discuss how the CI system and the Gate can deploy and test multi-node setuips.
Specifically, the TripleO team has been working on a heat-based installation of OpenStack that is also hopefully going to be referenced/used as part of the refstack work. This work allows us to do multi-node deploys without needing to pull in non-OpenStack deployment technology that may be contentious such as puppet, chef, juju or maas. Additionally, work has already started on integrating this work with OpenStack CI.
Talk about the approach in general, and what the status of CI integration is.