Attendees dazzled by spring skies still manage to dive into 80-minute working sessions to keep things humming along.


BARCELONA — Under beautiful blue Mediterranean skies, members of the OpenStack and Kubernetes communities once again demonstrated the power of open collaboration at KubeCon EU.

Setting the stage for the big show, the event kicked off with the contributor’s summit, where the OpenStack community was present in both the SIG-Cloud-Provider working session and in the SIG-Cluster-Lifecycle session. The Cloud Provider team continued working through the process of transitioning cloud-specific SIGs to working groups under the governance of SIG-Cloud-Provider. This aligns with the goal of removing cloud-specific code from upstream Kubernetes, while still giving cloud integrators a means of supporting their developers and users within the Kubernetes community.

Meanwhile, work continued on managing the complete life cycle of Kubernetes clusters on various cloud architectures using the Kubernetes API in the “Cluster Lifecycle” session. OpenStack’s representation in this work is strong, with a OpenStack implementation of the new Cluster API under active development. With the Cluster API implementation users will be able to manage the entire life cycle of Kubernetes on OpenStack with basic declarations of machines and nodes. While installing Kubernetes is the first goal, its flexible and generic API will allow for more advanced management features like upgrades, auto-scaling, and auto-healing, all in a generic way that still takes advantages of the OpenStack API. Adding to this exciting work is the introduction of Metal3 to the Kubernetes community. Metal3 uses the cluster API to deliver Kubernetes on bare metal using OpenStack Ironic.

The main event kicked off the next day, with several sessions devoted to cloud provider integrations. SIG-OpenStack community leaders including myself, Aditi Sharma and Christoph Glaubitz coordinated an 80-minute session that covered a diverse set of topics, including the OpenStack cloud provider, the OpenStack Cluster API provider and a cluster autoscaler for the Magnum cluster management project.

Working in collaboration with the cloud-provider community, SIG-Cloud-Provider met for another 80-minute session to cover shared goals. This included the primary goal of removing upstream provider code from Kubernetes (a note to Kubernetes users: if you’re depending on the upstream code, start migrating now to external providers, including OpenStack, before they disappear in late 2019!), code organization for sub-projects and the future of user and working groups for cloud providers.

On the last day, I teamed up with Fabio Rapposelli for a dive into the technical depths of Building a Controller Manager for cloud platforms, offering up tips and tricks drawn from the OpenStack provider.

About the author

Chris Hoge is strategic program manager at the OSF.

Photo // CC BY NC