Master Multi-Cluster Kubernetes with Kubefed: Cross-Cloud Scheduling & Failover

Overview

Kubernetes Federation (Kubefed) provides a native solution for managing multiple Kubernetes clusters as a single logical entity, enabling unified resource management, cross‑region application deployment, and automated failover for multi‑cloud or hybrid‑cloud strategies.

Key Technical Features

Unified Management UI : Single control plane to manage all clusters via a consistent API and CLI.

Intelligent Resource Scheduling : Supports multi‑dimensional policies based on region, capacity, and latency.

High‑Availability Failover : Automatic health checks and workload migration on cluster failure.

Cross‑Cloud Resource Orchestration : Abstracts differences between AWS, Azure, GCP, Alibaba Cloud, etc.

Progressive Rollout : Supports canary and blue‑green deployments across clusters.

Fine‑Grained RBAC : Role‑based access control across clusters for multi‑tenant isolation.

Typical Use Cases

Multi‑cloud active‑active architectures to avoid vendor lock‑in.

Cross‑region disaster recovery for finance, e‑commerce, and other high‑availability workloads.

Edge computing with hundreds of small clusters managed centrally.

Isolated environments for development, testing, pre‑release, and production.

Compliance‑driven data residency requirements.

Environment Requirements

Linux OS: CentOS 7+, Ubuntu 20.04+, Debian 10+

Kubernetes version: 1.23+ (both host and member clusters)

Kubefed version: v0.10.0+

kubectl version: 1.23+ (must match cluster version)

Helm 3.8+ (for installing Kubefed)

Network connectivity between clusters (VPN, dedicated line, or public internet)

Step‑by‑Step Implementation

1. Preparation

System Check

# Check OS version
cat /etc/os-release

# Verify each cluster
kubectl cluster-info
kubectl get nodes
kubectl version --short

# Check resource status
free -h
df -h

# Verify inter‑cluster network
curl -k https://cluster1.example.com:6443/healthz
curl -k https://cluster2.example.com:6443/healthz

Install Dependencies

# Install kubectl if missing
curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
sudo install -o root -g root -m 0755 kubectl /usr/local/bin/kubectl
kubectl version --client

# Install Helm 3
curl https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash
helm version

# Install kubefedctl CLI
wget https://github.com/kubernetes-sigs/kubefed/releases/download/v0.10.0/kubefedctl-0.10.0-linux-amd64.tgz
 tar -zxvf kubefedctl-0.10.0-linux-amd64.tgz
sudo mv kubefedctl /usr/local/bin/
kubefedctl version

# Verify installations
which kubectl kubefedctl helm

Prepare kubeconfig Files

# Create working directory
mkdir -p ~/kubefed-demo/kubeconfigs
cd ~/kubefed-demo

# Copy kubeconfig of each cluster
cp ~/.kube/config-host kubeconfigs/host-cluster.yaml
cp ~/.kube/config-cluster1 kubeconfigs/cluster1.yaml
cp ~/.kube/config-cluster2 kubeconfigs/cluster2.yaml

# Set host cluster context
export KUBECONFIG=~/kubefed-demo/kubeconfigs/host-cluster.yaml
kubectl config current-context
kubectl get nodes

2. Core Configuration

Deploy Kubefed Control Plane

# Create namespace for Kubefed on host cluster
kubectl create namespace kube-federation-system

# Add Helm repo and install Kubefed chart
helm repo add kubefed-charts https://raw.githubusercontent.com/kubernetes-sigs/kubefed/master/charts
helm repo update
helm install kubefed kubefed-charts/kubefed \
  --namespace kube-federation-system \
  --set controllermanager.replicaCount=2 \
  --set controllermanager.resources.requests.cpu=200m \
  --set controllermanager.resources.requests.memory=256Mi \
  --set controllermanager.resources.limits.cpu=500m \
  --set controllermanager.resources.limits.memory=512Mi

# Verify components
kubectl get pods -n kube-federation-system
kubectl get deployment -n kube-federation-system
kubectl get crd | grep kubefed

Note : Deploy the control plane on a dedicated host cluster and allocate at least two replicas for HA.

Register Member Clusters

# Register host cluster itself
kubefedctl join host-cluster \
  --cluster-context host-cluster \
  --host-cluster-context host-cluster \
  --kubefed-namespace kube-federation-system \
  --v=2

# Register cluster1
kubefedctl join cluster1 \
  --cluster-context cluster1 \
  --host-cluster-context host-cluster \
  --kubefed-namespace kube-federation-system \
  --kubeconfig ~/kubefed-demo/kubeconfigs/cluster1.yaml \
  --v=2

# Register cluster2
kubefedctl join cluster2 \
  --cluster-context cluster2 \
  --host-cluster-context host-cluster \
  --kubefed-namespace kube-federation-system \
  --kubeconfig ~/kubefed-demo/kubeconfigs/cluster2.yaml \
  --v=2

# Verify registration
kubectl get kubefedclusters -n kube-federation-system
kubectl describe kubefedcluster cluster1 -n kube-federation-system

# Check health status
kubectl get kubefedclusters -n kube-federation-system -o jsonpath='{range .items[*]}{.metadata.name}{