Purpose
This guide provides instructions to install Bare Metal Orchestrator and create worker nodes for an initial cluster deployment, as well as how to upgrade Bare Metal Orchestrator, and set up high availability.
Audience
This guide is primarily intended for administrators who are responsible to deploy and upgrade Bare Metal Orchestrator nodes.
Disclaimer
This guide may contain language that is not consistent with Dell Technologies’ current guidelines. Dell Technologies plans to update the guide over subsequent future releases to revise the language accordingly.
Terms used in this guide
The following terms are used in this guide:
- five-node high availability cluster
-
For a high availability (HA) Bare Metal Orchestrator cluster, the Global Controller (CP1) control data and services are fully replicated on two redundant HA nodes (CP2 and CP3). There is also a redundant pair of Load Balancers that provide highly reliable management access to Bare Metal Orchestrator. This makes up the five nodes in an HA Bare Metal Orchestrator configuration.
- Global Controller
- Bare Metal Orchestrator installs a single node RKE2 (next-generation) cluster called the Global Controller (GC) node that is also called control plane 1 (CP1) for high availability (HA) deployments. The Global Controller is sometimes called the GC site.
- kubeconfig file
- This YAML file is generated when a user is created in Bare Metal Orchestrator and is required for user authentication and cluster access. It contains the user's configuration details such as the username, token values, certificate data, and the IP address of the server to connect to access the cluster.
- Load Balancers
- In HA configurations, a redundant pair of Load Balancers manage load distribution to the three control plane nodes (CP1, CP2, and CP3). The virtual IP (VIP) address of the Load Balancers is used for CLI, Web user interface, and API access to the Bare Metal Orchestrator cluster.
- redundant HA nodes
- When deploying a high availability (HA) Bare Metal Orchestrator cluster, there are two HA control plane nodes (CP2 and CP3) that function as a redundant pair for failover in case the Global Controller (CP1) node fails.
- single node cluster
- You can deploy Bare Metal Orchestrator on physical servers or VMs as a single node cluster. A single node cluster has a single Global Controller node, to which you can add remote worker nodes and sites. You cannot convert a single node cluster to a high availability (HA) cluster.
Conventions used in this guide
Italicized content contained within angle brackets (< >) identify variables in CLI command strings. For example: https://<bmo-hostname>
, where <bmo-hostname>
is the variable. When entering the variable content, do not include the angle brackets. For example: https://bmo-manager-1
You can install Bare Metal Orchestrator on virtual machines (VMs) and physical servers. However, throughout this guide, we mention VMs specifically.