Creating the Cluster

Configurations

I created my nodes by making clones from an Cloud-init template of an Ubuntu server that I set up. To check out how I created a template with Ubuntu Cloud Images and cloud-init on Proxmox, check out my Proxmox Book.

After getting getting my Ubuntu Servers up and running, installing qemu-guest-agent, and upgrading all packages, I took a snapshot of the VM to have something to rollback to incase I mess up at any point. Taking snapshots is a great practice especially when learning new technologies, as it saves you time from recreating from scratch and allows you to roll back to certain states you save. The following are steps I took on my Master-Node to configure and prepare my nodes for my Kubernetes Cluster.

After establishing my SSH connection to my server, I followed the steps below and ran the corresponding commands to configure:

VM Set Up and Initial Configurations

Disable Swap

Kubernetes does not get along well with swap enabled, so I need to disable it. To do this, I ran the command in my terminal window to temporarily disable swap:

sudo swapoff -a

Next, I ran the command to edit the fstab file and comment out and keep swap turned off even after reboots:

sudo nano /etc/fstab

To confirm I've done this correctly, run the command to view swap usage:

free -m

Static IP or Static Leases

Your nodes will each require a static IP or Static DHCP lease. I've set up Static Leases for all of my nodes via OPNsense. To assign static IPs directly on the server, you'll need to edit and make changes to the .yaml file in the netplan directory.

Install Container Runtime (containerd)

Kubernetes requires a container runtime. I'll be using Containerd. To get started with this, run the following command:

I ran the command below to install containerd:

sudo apt install containerd

Check that the service is running by following command:

systemctl status containerd

After confirming it's running, create a new directory for containerd within /etc by running the following command:

sudo mkdir /etc/containerd

Next, write the default configuration to containerd by running the following command:

containerd config default | sudo tee /etc/containerd/config.toml

Now, use whatever editor you like, I use nano, and edit the config file. Search for runc.options within the file, and change the systemdCgroup value to true:

sudo nano /etc/containerd/config.toml

Enter Ctrl+W to search within the file for runc.options and change the SystemdCgroup value

Systemctl.conf Configuration

Run the following command to edit the sysctl.conf file:

sudo nano /etc/sysctl.conf

Find the line that enables packet forwarding for IPv4 and uncomment it so it can be read by the system:

This allows your nodes to communicate with each other and is crucial to get your cluster to work

k8s.conf Configuration

The next file to edit is the k8s.conf file. Run the following command:

sudo nano /etc/modules-load.d/k8s.conf

This is a blank file. Add "br_netfilter" and save the file:

The bridge netfilter essentially ensures that network bridging is supported throughout the cluster. After this is done. reboot your VMs.

 

Kubernetes Installation

With our VMs running, we can now install Kubernetes packages. To do this, we'll need to add the Kubernetes repository gpg key and then install the repository itself. Then we can run the commands to install kubeadm, kubelet, and kubectl.

Install Kubernetes using Native Package Management

Run the command to install necessary packages in order to use the Kubernetes apt packages:

sudo apt-get install -y apt-transport-https ca-certificates curl

Run the command to make sure you have the "keyrings" directory prior to downloading the public signing key:

ls -l /etc/apt

Download the public signing key for Kubernetes package repositories. I'm downloading the most current stable release v1.29:

curl -fsSL https://pkgs.k8s.io/core:/stable:/v1.29/deb/Release.key | sudo gpg --dearmor -o /etc/apt/keyrings/kubernetes-apt-keyring.gpg

Run the command to add the Kubernetes apt repositories:

echo 'deb [signed-by=/etc/apt/keyrings/kubernetes-apt-keyring.gpg] https://pkgs.k8s.io/core:/stable:/v1.29/deb/ /' | sudo tee /etc/apt/sources.list.d/kubernetes.list

Run the commands to update and upgrade apt repository and packages, and then install Kubernetes:

sudo apt-get update && sudo apt-get upgrade
sudo apt-get install -y kubectl kubeadm kubelet

Create Worker Template Node (optional)

At this stage. we now have a node that is properly configured and has the Kubernetes packages installed. I created a template of this VM so that in the future, I can add nodes to any cluster much faster and skip all of this configuration and initial set up. It's relatively simple to do this in Proxmox, but first, we'll want to clean up our VM so that static reservations and other configurations won't get cloned. Run the following commands to do so:

First, clean cloud-init with the following command:

sudo cloud-init clean

Remove the instances in the cloud repository by running the following command:

sudo rm -rf /var/lib/cloud/instances

Next, reset the machine-id (this avoids having your clones use the same static IP) by running the following command:

sudo truncate -s 0 /etc/machine-id

Remove the machine-id in the dbus directory by running the following command, and then create a symbolic link by running the following commands:

sudo rm /var/lib/dbus/machine-id
sudo ln -s /etc/machine-id /var/lib/dbus/machine-id

You can now power off this VM and and convert it to a template.

 

Initialize Pod Network on your Controller Node

From my template, I've created 4 clones:

• k-ctrlr
• k8-n1
• k8-n2
• k8-n3

My cluster will have 1 Controller node, and 3 worker nodes. The specs for my nodes are as follows:

• Controller
  • 4 Gb RAM
  • 4 CPU Cores
• Worker
  • 2 GB RAM
  • 2 CPU Cores

After powering up the node vms, ensure they have static IPs, or assign Static leases in your router (OPNsense in my case). 

To initialize the following command, run the following command after editing certain parameters:

sudo kubeadm init --control-plane-endpoint=192.168.2.39 --node-name k-ctrlr --pod-network-cidr=10.244.0.0/16

• For --control-plan-endpoint, edit this to make it your Controller Nodes IP. In my case, I've assigned k-ctrlr 192.168.2.39
• For --node-name, enter your Controller Node's hostname, in my case it is k-ctrlr
• Leave pod network the same. Changing this is possible, but will require additional configurations. 

 

Save

After Keysinitialization, you'll see the following output, which contains commands including keys and tokens to Joinadd Nodes

nodes to the cluster:

• I've greyed out my tokens and hash value. Save these in a .txt file or somewhere accessible as we'll be using this in the near future to join our other 3 nodes to the cluster.

To complete this initialization and to allow yourself to control your cluster as a regular user run the following commands:

mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

After doing this, your cluster has been initialized and is almost ready to add nodes. 

Overlay Network

After you initialize your Controller node, run the command to view your pods:

 

Adding Nodes to Cluster

Nodeport Service for Internet Access