Skip to content

Latest commit

 

History

History
90 lines (68 loc) · 3.75 KB

README.md

File metadata and controls

90 lines (68 loc) · 3.75 KB
Raw

K3s

K3s is a terrific Kubernetes distribution from Rancher Labs which uses 40MB worth of system resources - ideal for edge deployments. But it's also perfect for running full-scale Kubernetes clusters due to its simplicity.

This playbook installs a working K3s cluster with a single master node and multiple worker nodes running Ubuntu 18.04, using a deployment node running CentOS 7.

Prerequisites

  1. A deployment node where this playbook will run.
  2. Nodes for hosting K3s master and worker components with root access over SSH.

Installation

  1. Clone this repository and install its dependencies:
git clone https://github.com/rudimk/eurus-fabric-playbooks.git
cd eurus-fabric-playbooks
pip install -r requirements.txt
  1. On the deployment node, update yum and install Python 3.6:
yum update
yum install centos-release-scl
yum install rh-python36
  1. To activate Python 3, you need to use scl to activate a new shell which uses Python 3 instead of the default Python 2 installation(yeah, CentOS is fucked up). Run the command below whenever you need to run the playbook(you could also put it in your .bashrc or .bash_profile):
scl enable rh-python36 bash

  1. Ensure that the nodes you intend to deploy K3s on are accessible from the deployment node. Also make sure you can log into these nodes over SSH without a password. Here's a quick primer on creating a new keypair and setting up password-less SSH auth:

    a. Create a new keypair without a passphrase:

    ssh-keygen -t rsa -b 4096 -C "your_email@example.com"

    Try giving a different name for this particular keypair, to differentiate it from other keypairs. Here, we'll use id_rsa_k3s as an example.

    b. Create a SSH config file at ~/.ssh/config with the following entries(one entry per K3s node):

    Host <NODE_IP>
	IdentityFile ~/.ssh/id_rsa_k3s

    c. One disadvantage with running Python 3 inside a separate shell with scl is that you need to manually add the SSH keypair to the SSH agent on the deployment node. To take care of that:

    eval "$(ssh-agent -s)" && ssh-add ~/.ssh/id_rsa_k3s

  2. Finally, it's time to populate the inventory file. Open k3s/ubuntu18/inventory.py and fill in the following details:

    a. Provide the hostname and IP address of the master node in K3S_MASTER_NODE: K3S_MASTER_NODE = {'hostname': 'k3s-master', 'ip': '10.1.0.1'}.

    b. Provide hostnames and IP addresses for your worker nodes:

    K3S_WORKER_NODES = [{'hostname': 'k3s-worker-1', 'ip': '10.1.0.2'}, {'hostname': 'k3s-worker-2', 'ip': '10.1.0.3'}]

    c. Finally, define a list of hosts that will be injected into the /etc/hosts file on the deployment machine as well as on all K3s nodes. Here's a sample:

    
10.1.0.1	k3s-master
10.1.0.2	k3s-worker-1
10.1.0.3	k3s-worker-2

    Ensure that the hostnames and IPs mentioned here match those in K3S_MASTER_NODE and K3S_WORKER_NODES, as well as the actual hostnames/IPs of the nodes you're planning to use.

  3. Now you're all set. Navigate to the k3s/ubuntu18 directory and fire up main.py:

cd k3s/ubuntu18
python main.py
  1. The installer will deploy a K3s cluster on your nodes, and download a kubeconfig.yaml file to /root on your deployment machine. Modify the server variable in the file to https://<K3S_MASTER_HOSTNAME>:6443. Install kubectl and configure it to interact with your new cluster:
yum install -y kubectl
mkdir -p /root/.kube
cp /root/kubeconfig.yaml /root/.kube/config
  1. Verify your cluster is operational and reachable:
kubectl get nodes

Next Steps:

  1. Persistent Storage with Ceph: Manifests