Documentation of a mini bare-metal cluster intended for Kubernetes experiments. It consists in 6 nodes (X86_64 architecture) managed by a server node providing global services (DNS, etc.). The whole cluster fits in a case and is therefore portable.
Day-to-day information on the cluster is available on my blog.
All information related to hardware is gathered in a dedicated page.
In addition to providing access to the minicluster, the gateway also hosts global services and network configuration for cluster nodes. A dedicated page gathers information on gateway configuration.
As much as possible, and once public keys of management accounts have been deployed on nodes, global configuration is done using Ansible scripts
Initial plan was to deploy CentOS stream using provisioning tools like The Foreman for nodes life cycle management. As I discovered that cluster nodes couldn't boot with PXE, a more traditional and low level approach had to be used.
Furthermore, due to change in CentOS project management, I switched to Rocky Linux alternative.
Automated deployment of Operating System on cluster nodes is documented on a dedicated page.
Documentation available soon: stay tuned.
Documentation available soon: stay tuned.
- Single cluster
- Two clusters in separate VLANs