This playbook provisions a CentOS 7.3/Ubuntu 16.04/RHEL/Fedora 25+ machine with everything you need in order to
use kubeadm
and CRI-O
(no docker
anymore). Simply run:
Create an ansible hosts
file (adjust anything needed like ansible_python_interpreter
):
$ cat hosts
192.168.1.122 ansible_python_interpreter='python3'
192.168.1.123
make sure the machine is reachable via SSH w/o password input before running the playbook
ssh-copy-id <user>@<ip or hostname>
Then simply run:
$ ansible-playbook -i hosts cri-o.yml -u <user>
This playbook installs kubernetes software so you can follow the kubeadm
docs to setup a Kubernetes cluster. I've tested the whole flow with a custom Flannel pod network, the official one seems broken.
cri-o-ansible
is not going to install any CNI network configuration. It just installs CNI plugins. Network configuration is left to Kubernetes and Openshift (or you following this)