Kubernetes is a popular platform for deploying, scaling and managing microservices due to its ease of use and application centric architecture. However, if you are looking to kick the tires with docker+kubernetes on OpenStack using nova instances and neutron overlay networking, you will soon find that there are very few resources available in terms of docs, automation and code.We realized this issue when we started exploring for resources to build our private kube cluster for one of our internal POCs. Eventhough the Magnum service is available, we did not use it in our test environment as it lacked hybrid cloud and many of our customers preferred deploying docker+kubernetes natively without the management complexity of Magnum. We have since spent time developing the required automation in ansible to launch repeatable kubernetes clusters in openstack from a cloud yaml spec. This talk will the present the details of the work we have done and lessons learned and strategies employed.
Attendees can learn the automation patterns we have used, issues that we have run into and the strategies we have employed.