Kubernetes Cluster Overview¶

K3s Distribution¶

The cluster runs K3s, a lightweight certified Kubernetes distribution optimized for resource-constrained environments.

Version Information¶

K3s Version: v1.34.3
Container Runtime: containerd
CNI: Flannel (VXLAN)

Cluster Architecture¶

graph TB
    subgraph Control Plane
        API[API Server]
        ETCD[etcd]
        CM[Controller Manager]
        Sched[Scheduler]
    end

    subgraph Worker Nodes
        W1[Worker 1]
        W2[Worker 2]
        W3[Worker 3]
        W4[Worker 4]
    end

    API --> W1
    API --> W2
    API --> W3
    API --> W4

    subgraph Per Node
        Kubelet[kubelet]
        KubeProxy[kube-proxy]
        Containerd[containerd]
    end

Node Status¶

Node	Role	Status
master	control-plane	Ready
worker-1	worker	Ready
worker-2	worker	Ready
worker-3	worker	Ready
worker-4	worker	Ready

Installed Components¶

Core K3s Components¶

CoreDNS: Cluster DNS
Traefik: Ingress controller (disabled, using Cloudflare Tunnel)
Local Path Provisioner: Basic storage (supplemented by Longhorn)
Metrics Server: Resource metrics

Add-ons¶

Component	Namespace	Purpose
ArgoCD	argocd	GitOps continuous deployment
Longhorn	longhorn-system	Distributed block storage
Prometheus	monitoring	Metrics collection
Grafana	monitoring	Visualization
Elasticsearch	monitoring	Log storage
Kibana	monitoring	Log analysis

Cluster Access¶

kubectl Configuration¶

# Get kubeconfig from master
export KUBECONFIG=/etc/rancher/k3s/k3s.yaml

ArgoCD CLI¶

# Login to ArgoCD
argocd login argocd.ajandrews.pro

Resource Quotas¶

Default resource limits per namespace:

apiVersion: v1
kind: ResourceQuota
metadata:
  name: default-quota
spec:
  hard:
    requests.cpu: "4"
    requests.memory: 8Gi
    limits.cpu: "8"
    limits.memory: 16Gi

Health Checks¶

Cluster Health¶

# Check node status
kubectl get nodes

# Check system pods
kubectl get pods -n kube-system

# Check component status
kubectl get componentstatuses

Common Issues¶

Issue	Symptom	Resolution
Node NotReady	Node shows NotReady	Check kubelet, network
Pod Pending	Pods stuck pending	Check resources, node selector
ImagePullBackOff	Container can't pull	Check registry credentials

Upgrades¶

K3s Upgrade Process¶

Backup etcd data
Upgrade master node
Upgrade worker nodes (rolling)
Verify all nodes Ready

# On master
curl -sfL https://get.k3s.io | sh -

# On workers
curl -sfL https://get.k3s.io | K3S_URL=https://master:6443 K3S_TOKEN=<token> sh -