Guide to Installing OpenShift 4.10 IPI on Bare Metal

If you are starting your journey with OpenShift in a lab environment, I recommend using the Assisted Installer with my two-part guide and the accompanying YouTube videos, instead of this guide.

This guide will cover installing OpenShift 4.10 on bare metal using the IPI installer (installer-provisioned infrastructure). The IPI method can be described as a fully automated install that uses the OpenShift installer program to provision the cluster. In this guide, using the information we provide in the install-config.yaml, the installer program will power on our three bare metal servers via IPMI (Intelligent Platform Management Interface) and complete the cluster installation in a fully automated fashion.

One challenging portion of this guide will be following how my network is designed and then adapting your network to match the cluster requirements. For example, your servers, IP addresses, DNS, DHCP, and MAC addresses will differ from what I use in this guide. About 75% of the issues mentioned in the comments of previous guides are networking-related and typically where you want to look first if you run into problems during the installation. Let’s get started.

Network Layout

Network Hardware

I use a Ubquiti Dream Machine as my router/gateway and a 24-port Ubquiti switch. We need one “baremetal” network for this IPI installation and one “Provisioning” network. Ports for the iDRAC and NIC2 are connected to my “Default Network” profile and are considered the “baremetal” network for this guide. The NIC1 on each server is connected to the “Provisioning Network” on ports 22–24 of the switch and set to VLAN 10.

DNS Setup

For local DNS, I utilize pi-hole installed on a VM and a BegleBoneBlack.