Install the tools. Here’s the setup for a Fedora/RHEL box.
mkdir ~/bin
wget -O /tmp/ocp/openshift-client-linux.tar.gz https://mirror.openshift.com/pub/openshift-v4/x86_64/clients/ocp/stable/openshift-client-linux.tar.gz
tar -xvf /tmp/ocp/openshift-client-linux.tar.gz -C ~/bin
wget -O /tmp/ocp/openshift-install-linux.tar.gz https://mirror.openshift.com/pub/openshift-v4/x86_64/clients/ocp/stable/openshift-install-linux.tar.gz
tar -xvf /tmp/ocp/openshift-install-linux.tar.gz -C ~/bin
sudo dnf install /usr/bin/nmstatectl -yInstall the Cluster
One of the things about the OpenShift agent based installer is that when you provide it configs, it actually destroys them. The way I get around this, if I am iterating in my installs, is to do the following.
rm -rf install
mkdir install
cp agent-config.yaml install-config.yaml install
openshift-install agent create image --dir=install --log-level=debugThen once you boot the machines with the iso created, then run the following commands to complete the install.
openshift-install agent wait-for bootstrap-complete --dir=install --log-level=debug
...and when that is complete...
openshift-install agent wait-for install-complete --dir=install --log-level=debugConfig Files
Here’s the configs for my homelab setup, which is a bare-metal 3 node cluster on some Intel NUCs.
apiVersion: v1
baseDomain: lab.snimmo.com
metadata:
name: ocp
compute:
- architecture: amd64
hyperthreading: Enabled
name: worker
replicas: 0
controlPlane:
architecture: amd64
hyperthreading: Enabled
name: master
replicas: 3
networking:
clusterNetwork:
- cidr: 10.128.0.0/14
hostPrefix: 23
machineNetwork:
- cidr: 10.3.0.0/24
networkType: OVNKubernetes
serviceNetwork:
- 172.30.0.0/16
platform:
baremetal:
apiVIP: 10.3.0.9
ingressVIP: 10.3.0.10
pullSecret: '<redacted>'
sshKey: 'ssh-ed25519 AAAA....ZcOs'And the agent based install.
apiVersion: v1alpha1
kind: AgentConfig
metadata:
name: ocp
rendezvousIP: 10.3.0.11
hosts:
- hostname: nuc1
role: master
interfaces:
- name: enp57s0u1c2
macAddress: F8:E4:3B:BE:2E:16
rootDeviceHints:
deviceName: "/dev/sda"
networkConfig:
interfaces:
- name: enp57s0u1c2
type: ethernet
state: up
mac-address: F8:E4:3B:BE:2E:16
ipv4:
enabled: true
address:
- ip: 10.3.0.11
prefix-length: 24
dhcp: false
dns-resolver:
config:
server:
- 10.3.0.2
routes:
config:
- destination: 0.0.0.0/0
next-hop-address: 10.3.0.1
next-hop-interface: enp57s0u1c2
table-id: 254
- hostname: nuc2
role: master
interfaces:
- name: eno1
macAddress: 1C:69:7A:61:5E:AE
rootDeviceHints:
deviceName: "/dev/sda"
networkConfig:
interfaces:
- name: eno1
type: ethernet
state: up
mac-address: 1C:69:7A:61:5E:AE
ipv4:
enabled: true
address:
- ip: 10.3.0.12
prefix-length: 24
dhcp: false
dns-resolver:
config:
server:
- 10.3.0.2
routes:
config:
- destination: 0.0.0.0/0
next-hop-address: 10.3.0.1
next-hop-interface: eno1
table-id: 254
- hostname: nuc3
role: master
interfaces:
- name: eno1
macAddress: 1C:69:7A:61:64:F4
rootDeviceHints:
deviceName: "/dev/sda"
networkConfig:
interfaces:
- name: eno1
type: ethernet
state: up
mac-address: 1C:69:7A:61:64:F4
ipv4:
enabled: true
address:
- ip: 10.3.0.13
prefix-length: 24
dhcp: false
dns-resolver:
config:
server:
- 10.3.0.2
routes:
config:
- destination: 0.0.0.0/0
next-hop-address: 10.3.0.1
next-hop-interface: eno1
table-id: 254Alternate Install – Bond with Single NIC
In the following agent config file, I wanted to create a bit more of an “enterprisey” setup, so I created a bond interface, which is active-passive, but with only a single ethernet interface backing it up because my NUCs don’t have dual NICs. More for documentation and to see if it would work.
apiVersion: v1alpha1
kind: AgentConfig
metadata:
name: ocp
rendezvousIP: 10.3.0.11
hosts:
- hostname: nuc1
role: master
interfaces:
- name: enp57s0u1c2
macAddress: F8:E4:3B:BE:2E:16
rootDeviceHints:
deviceName: "/dev/sda"
networkConfig:
interfaces:
- name: enp57s0u1c2
type: ethernet
state: up
mac-address: F8:E4:3B:BE:2E:16
- name: bond0
type: bond
state: up
ipv4:
enabled: true
address:
- ip: 10.3.0.11
prefix-length: 24
dhcp: false
link-aggregation:
mode: active-backup
port:
- enp57s0u1c2
options:
miimon: '100'
primary: enp57s0u1c2
dns-resolver:
config:
server:
- 10.3.0.2
routes:
config:
- destination: 0.0.0.0/0
next-hop-address: 10.3.0.1
next-hop-interface: bond0
table-id: 254
- hostname: nuc2
role: master
interfaces:
- name: eno1
macAddress: 1C:69:7A:61:5E:AE
rootDeviceHints:
deviceName: "/dev/sda"
networkConfig:
interfaces:
- name: eno1
type: ethernet
state: up
mac-address: 1C:69:7A:61:5E:AE
- name: bond0
type: bond
state: up
ipv4:
enabled: true
address:
- ip: 10.3.0.12
prefix-length: 24
dhcp: false
link-aggregation:
mode: active-backup
port:
- eno1
options:
miimon: '100'
primary: eno1
dns-resolver:
config:
server:
- 10.3.0.2
routes:
config:
- destination: 0.0.0.0/0
next-hop-address: 10.3.0.1
next-hop-interface: bond0
table-id: 254
- hostname: nuc3
role: master
interfaces:
- name: eno1
macAddress: 1C:69:7A:61:64:F4
rootDeviceHints:
deviceName: "/dev/sda"
networkConfig:
interfaces:
- name: eno1
type: ethernet
state: up
mac-address: 1C:69:7A:61:64:F4
- name: bond0
type: bond
state: up
ipv4:
enabled: true
address:
- ip: 10.3.0.13
prefix-length: 24
dhcp: false
link-aggregation:
mode: active-backup
port:
- eno1
options:
miimon: '100'
primary: eno1
dns-resolver:
config:
server:
- 10.3.0.2
routes:
config:
- destination: 0.0.0.0/0
next-hop-address: 10.3.0.1
next-hop-interface: bond0
table-id: 254Here’s what the network interfaces look like on the host after this type of install.
[core@nuc2 ~]$ nmcli c sh
NAME UUID TYPE DEVICE
ovs-if-br-ex fd87a8ba-8d32-4e76-bc14-505cf66ebc51 ovs-interface br-ex
br-ex b7786e2d-0da2-4903-909e-7e4ffa627402 ovs-bridge br-ex
eno1-slave-ovs-clone dd89feca-823f-4d61-8866-109c8fd68f7e ethernet eno1
ovs-if-phys0 6193d237-4d3f-43b4-9040-bb0675896215 bond bond0
ovs-port-br-ex 07fb551e-af0d-4e48-95cd-4874afa87ac7 ovs-port br-ex
ovs-port-phys0 d8e8f512-91af-4404-a6df-2c4a85e85115 ovs-port bond0
lo b9378fb3-1369-4ac7-ada1-1f62564637b1 loopback lo
bond0 925b4a95-2de0-5b2d-bcf5-8b684a7e9cb4 bond --
eno1 ff8b47e2-9b1c-5e04-9f8b-34671372837c ethernet --
2 comments
Comments are closed.