Revisiting OpenShift Installation Automation: From Custom ISOs to Modern Methods

Generated by ChatGPT

When OpenShift 4 debuted with its Installer-Provisioned Infrastructure (IPI), many of us developed creative solutions to streamline deployments. In my early experience with 4.x versions, I approached automated installations through ISO customization — a method that served its purpose but more like a hack.

In essence, for the bare-metal/VM deployments, I modified the original RHCOS installation media by updating its isolinux.cfg to embed the kernel parameters like below (joining them together as one line)

rd.neednet=1
ip={{ .ip }}::{{ .gateway }}:{{ .mask }}:{{ .hostname }}:{{ .iface }}:none
nameserver={{ .nameserver }}
coreos.inst.install_dev={{ .diskName }}
coreos.inst.ignition_url={{ .ignUrl }}

Based on this, I created the custom ISO file for each of the VM and let it boot up without manual intervention. It worked fine though I need to validate the changes for each release. I recalled after some version the installation started to validate the CD label so that I need to add labeling for my custom ISO file and make sure it match in the isolinux.cfg file.

Since 4.11 agent-base installation is introduced till now its getting more mature. Time to evolve. Let’s explore the agent-based OpenShift installation on KVM with version 4.16.