giannisalinetti/libvirt-ocp4-provisioner

Automate your OCP4 installation

libvirt-ocp4-provisioner - Automate your cluster provisioning from 0 to OCP!

Welcome to the home of the project! This project has been inspired by @ValentinoUberti, who did a GREAT job creating the playbooks to provision existing infrastructure nodes on oVirt and preparing for cluster installation.

I wanted to play around with terraform and port his great work to libvirt and so, here we are! I adapted his playbooks to libvirt needs, making massive use of in-memory inventory creation for provisioned VMs, to minimize the impact on customizable stuff in variables.

To give a quick overview, this project will allow you to provision a fully working OCP stable environment, consisting of:

• Bastion machine provisioned with:
  • dnsmasq (with SELinux module, compiled and activated)
  • dhcp based on dnsmasq
  • nginx (for ignition files and rhcos pxe-boot)
  • pxeboot
• Loadbalancer machine provisioned with:
  • haproxy
• OCP Bootstrap machine
• OCP Master(s) VM(s)
• OCP Worker(s) VM(s)

It also takes care of preparing the host machine with needed packages, configuring:

• dedicated libvirt network (fully customizable)
• dedicated libvirt storage pool (fully customizable)
• terraform
• libvirt-terraform-provider ( compiled and initialized based on https://github.com/dmacvicar/terraform-provider-libvirt)

PXE is automatic, based on MAC binding to different OCP nodes role, so no need of choosing it from the menus, this means you can just run the playbook, take a beer and have your fully running OCP 4.4.latest stable up and running.

bastion and loadbalancer VMs spec:

• OS: Centos8 Generic Cloud base image https://cloud.centos.org/centos/8/x86_64/images/
• cloud-init:
  • user: ocpinstall
  • pass: ocprocks
  • ssh-key: generated during vm-provisioning and stores in the project folder

The user is capable of logging via SSH too.

Quickstart

The playbook is meant to be ran against a/many local or remote host/s, defined under vm_host group, depending on how many clusters you want to configure at once.

ansible-playbook main.yml

You can quickly make it work by configuring the needed vars, but you can go straight with the defaults!

vars/libvirt.yml

libvirt:                       
  storage:                     
    pool_name: ocp4
    pool_path: /var/lib/libvirt/images/ocp4
  network:                     
    network_name: ocp4         
    network_gateway: 192.168.100.1
    network_bridge: ocpbr1     
    network_mask: 255.255.255.0

The kind of network created is a simple NAT configuration, without DHCP since it will be provisioned with bastion VM. Defaults can be OK if you don't have any overlapping network.

vars/infra_nodes.yml

domain: hetzner.lab
cluster_name: ocp4
nfs_registry: false
infra_nodes:
  host_list:
    bastion:
      - ip: 192.168.100.4
    loadbalancer:
      - ip: 192.168.100.5
dhcp:
  timezone: "Europe/Rome"
  ntp: 204.11.201.10

Where domain is the dns domain assigned to the nodes and cluster_name is the name chosen for our OCP cluster installation.

The variable nfs_registry is set to false by default. If set to true, it will deploy an additional 100Gi volume on bastion VM, create the PV and patch registry to use it in Managed mode.

vars/cluster_nodes.yml

cluster_nodes:
  host_list:
    bootstrap:
      - ip: 192.168.100.6
    masters:
      - ip: 192.168.100.7
      - ip: 192.168.100.8
      - ip: 192.168.100.9
    workers:
      - ip: 192.168.100.10
        role: infra
      - ip: 192.168.100.11
      - ip: 192.168.100.12
  specs:
    bootstrap:
      vcpu: 4
      mem: 16
      disk: 40
    masters:
      vcpu: 4
      mem: 16
      disk: 40	  
    workers:
      vcpu: 2
      mem: 8
      disk: 40
        
cluster:
  ocp_user: admin
  ocp_pass: openshift
  pullSecret: ''

mem and disk are intended in GB

The role for workers is intended for nodes labelling. Omitting labels sets them to their default value, worker

The count of VMs is taken by the elements of the list, in this example, we got:

• 3 master nodes with 4vcpu and 16G memory
• 3 worker nodes with 2vcpu and 8G memory

Recommended values are:

Role	vCPU	RAM	Storage
bootstrap	4	16G	120G
master	4	16G	120G
worker	2	8G	120G

For testing purposes, minimum storage value is set at 40GB.

Pull Secret can be retrived easily at https://cloud.redhat.com/openshift/install/pull-secret

HTPasswd provider is created after the installation, you can use ocp_user and ocp_pass to login!

DISCLAIMER This project is for testing/lab only, it is not supported in any way by Red Hat nor endorsed.

Feel free to suggest modifications/improvements.

Alex