How I setup my first Kubernetes Cluster with K0s
Intro With the fastest era of the tech industry, I think every software engineer has to catch up with this knowledge. Thats why I kicked off my project to learn K8s. The first stack I chose was K0s, recommended by my ex-leader, who’s super clever and talented. Please keep in mind, this isn’t a guidebook — it’s my diary, so don’t expect too much. No more rambling, lets dive in! Preparation I didn’t wanna run it on my local machine, so I decided to install it on my VPS. It’s finally time to use it after letting it collect dust for ages Make sure those port are accessible. If you have a firewall enabled, you’ll need to allow these ports # Control plane sudo ufw allow 6443/tcp # Kubernetes API server sudo ufw allow 8132/tcp # k0s admin API # Node communication sudo ufw allow 10250/tcp # kubelet API sudo ufw allow 10256/tcp # kube-proxy sudo ufw allow 2379:2380/tcp # etcd client/peer sudo ufw allow 179/tcp # BGP (Calico/MetalLB) sudo ufw allow 30000:32767/tcp # NodePort range # Ingress / HTTP(S) sudo ufw allow 80/tcp sudo ufw allow 443/tcp # DNS lookup outbound sudo ufw allow out 53/tcp sudo ufw allow out 53/udp Seek the flame 🔥 K0s Kick your as* into reading the K0s docs first before taking any next steps. Download latest k0s curl --proto '=https' --tlsv1.2 -sSf https://get.k0s.sh | sudo sh sudo curl -sSLf https://get.k0s.sh | sudo sh Install controller plane sudo k0s install controller --enable-worker --no-taints --no-tains removes the default taint from the control plane, allowing the scheduler to place regular Pods on this node. Start k0s sudo k0s start However, if you want to access the cluster from somewhere else (or use an independent install of kubectl), you’ll need the KUBECONFIG file. When you create the server, k0s automatically generates a KUBECONFIG file for you, so just copy it to your working directory and point to it. Configure KUBECONFIG sudo cp /var/lib/k0s/pki/admin.conf ./admin.conf export KUBECONFIG=./admin.conf Ok! Looks cool now! Configure kube-proxy for MetalLB compatibility k0s doesn’t come with a built-in load balancer, so I have to set up a real one myself. I decided to go with MetalLB Enable IPVS mode and strictARP export EDITOR=nano kubectl edit configmap -n kube-system kube-proxy Change content to: apiVersion: v1 data: config.conf: |- kind: KubeProxyConfiguration mode: "ipvs" ipvs: {...,"strictARP":true} Restart kube-proxy pods: kubectl delete pod -n kube-system -l k8s-app=kube-proxy Make sure all Pods in the kube-system namespace are running kubectl get pod -n kube-system Install MetalLB Go apply metallb manifest kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.15.2/config/manifests/metallb-native.yaml kubectl get pods -n metallb-system If MetalLB fails to create the memberlist secret automatically, create it manually kubectl create secret generic -n metallb-system memberlist --from-literal=secretkey="$(openssl rand -base64 128)" If you accidentally applied MetalLB before setting up kube-proxy, do a rollout restart: # Check which resources exist kubectl get deployments,daemonsets -n metallb-system # If there’s a Deployment kubectl rollout restart deployment/controller -n metallb-system # If there’s a DaemonSet kubectl rollout restart daemonset/speaker -n metallb-system Configure IP Address Pool Create a configuration file for MetalLB: metallb-config.yaml apiVersion: metallb.io/v1beta1 kind: IPAddressPool metadata: name: first-pool namespace: metallb-system spec: addresses: - YOUR_VPS_PUBLIC_IP/32 # cause VPS have only 1 Public IP so using /32 for subnet mask 255.255.255.255 --- apiVersion: metallb.io/v1beta1 kind: L2Advertisement metadata: name: example namespace: metallb-system spec: ipAddressPools: - first-pool Remember to replace YOUR_VPS_PUBLIC_IP IPAddressPool: Defines the range of IP addresses that MetalLB can allocate L2Advertisement: Configures Layer 2 mode to announce IPs externally ipAddressPools: Links the L2Advertisement to specific IP pools Apply configuration kubectl apply -f metallb-config.yaml Verify configuration # for ipaddresspool kubectl get ipaddresspool -n metallb-system # for l2advertisement kubectl get l2advertisement -n metallb-system Ok, done with the LB, the last thing is the Ingress Controller. Maybe I’ll go with Traefik instead of Nginx Ingress Controller. That’ll be in the next post. Now, I’m heading back to the bonfire 🔥 ...