# 下载集群联邦工具
$ git clone https://github.com/kubernetes-sigs/kubefed.git
$ cd kubefed
$ git checkout v0.8.1

# 构建kubefedctl工具
$ make build

# 安装集群联邦
$ export KUBECONFIG=~/.kube/config
$ kubefedctl init fed –host-cluster-context=cluster1 –addons-repo=https://github.com/kubernetes-sigs/kubefed/tree/master/addons

# 验证集群联邦安装
$ kubectl get pods -n kube-federation-system
NAME READY STATUS RESTARTS AGE
kubefed-admission-webhook-567890abc-12345 1/1 Running 0 5m
kubefed-controller-manager-567890abc-67890 1/1 Running 0 5m

# 查看集群联邦状态
$ kubectl get federatedclusters -n kube-federation-system
No resources found in kube-federation-system namespace.

# 查看可用的集群上下文
$ kubectl config get-contexts
CURRENT NAME CLUSTER AUTHINFO NAMESPACE
* cluster1 cluster1 user1 default
cluster2 cluster2 user2 default
cluster3 cluster3 user3 default

# 加入cluster2到联邦
$ kubefedctl join cluster2 –cluster-context=cluster2 –host-cluster-context=cluster1

# 加入cluster3到联邦
$ kubefedctl join cluster3 –cluster-context=cluster3 –host-cluster-context=cluster1

# 验证集群加入
$ kubectl get federatedclusters -n kube-federation-system
NAME AGE
cluster2 2m
cluster3 1m

# 查看集群状态
$ kubectl get federatedclusters -n kube-federation-system -o wide
NAME READY AGE
cluster2 True 3m
cluster3 True 2m

# 创建联邦命名空间
$ kubectl create namespace fgedu-app
$ kubefedctl federate namespace fgedu-app

# 部署联邦Deployment
$ kubectl apply -f – << 'EOF' apiVersion: types.federation.k8s.io/v1beta1 kind: FederatedDeployment metadata: name: webapp namespace: fgedu-app spec: template: metadata: labels: app: webapp spec: replicas: 6 selector: matchLabels: app: webapp template: metadata: labels: app: webapp spec: containers: - name: webapp 学习交流加群风哥微信: itpux-com image: harbor.fgedu.net.cn/library/webapp:v1.0.0 ports: - containerPort: 80 resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" placement: clusters: - name: cluster1 - name: cluster2 - name: cluster3 overrides: - clusterName: cluster1 clusterOverrides: - path: "/spec/replicas" value: 3 - clusterName: cluster2 clusterOverrides: - path: "/spec/replicas" value: 2 - clusterName: cluster3 clusterOverrides: - path: "/spec/replicas" value: 1 EOF # 部署联邦Service $ kubectl apply -f - << 'EOF' apiVersion: types.federation.k8s.io/v1beta1 kind: FederatedService metadata: name: webapp namespace: fgedu-app spec: template: metadata: labels: app: webapp spec: selector: app: webapp ports: - port: 80 targetPort: 80 placement: clusters: - name: cluster1 - name: cluster2 - name: cluster3 EOF # 查看联邦资源 $ kubectl get federateddeployments -n fgedu-app NAME AGE webapp 5m $ kubectl get federatedservices -n fgedu-app NAME AGE webapp 3m # 查看各集群中的资源 $ kubectl --context=cluster1 get deployments -n fgedu-app NAME READY UP-TO-DATE AVAILABLE AGE webapp 3/3 3 3 5m $ kubectl --context=cluster2 get deployments -n fgedu-app NAME READY UP-TO-DATE AVAILABLE AGE webapp 2/2 2 2 5m $ kubectl --context=cluster3 get deployments -n fgedu-app NAME READY UP-TO-DATE AVAILABLE AGE webapp 1/1 1 1 5m

# 部署联邦Ingress
$ kubectl apply -f – << 'EOF' apiVersion: types.federation.k8s.io/v1beta1 kind: FederatedIngress metadata: name: webapp namespace: fgedu-app spec: template: metadata: annotations: kubernetes.io/ingress.class: "nginx" spec: rules: - host: webapp.fgedu.net.cn http: paths: - path: / backend: serviceName: webapp servicePort: 80 placement: clusters: - name: cluster1 - name: cluster2 - name: cluster3 EOF # 配置DNS服务 $ kubectl apply -f - << 'EOF' apiVersion: types.federation.k8s.io/v1beta1 kind: FederatedDNSRecord metadata: name: webapp namespace: fgedu-app spec: template: spec: recordType: A ttl: 300 targets: - 192.168.1.100 - 192.168.2.100 - 192.168.3.100 placement: clusters: - name: cluster1 - name: cluster2 - name: cluster3 EOF

# 部署多集群负载均衡器
$ kubectl apply -f – << 'EOF' apiVersion: apps/v1 kind: Deployment metadata: name: multi-cluster-lb namespace: kube-system spec: replicas: 3 selector: matchLabels: app: multi-cluster-lb template: metadata: labels: app: multi-cluster-lb spec: containers: - name: multi-cluster-lb image: harbor.fgedu.net.cn/library/multi-cluster-lb:v1.0.0 ports: - containerPort: 80 env: - name: CLUSTER1_URL value: "http://192.168.1.100" - name: CLUSTER2_URL value: "http://192.168.2.100" - name: CLUSTER3_URL value: "http://192.168.3.100" resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" --- apiVersion: v1 kind: Service metadata: name: multi-cluster-lb namespace: kube-system spec: selector: app: multi-cluster-lb ports: - port: 80 targetPort: 80 type: LoadBalancer EOF # 查看负载均衡器状态 $ kubectl get services -n kube-system | grep multi-cluster-lb multi-cluster-lb LoadBalancer 10.96.78.90 192.168.0.100 80:30880/TCP 5m

。

Part04-生产案例与实战讲解

4.1 企业级多集群管理案例

某企业的多集群管理实践如下：

• 集群分布：在全球3个地区部署了5个K8s集群，包括生产、预生产和测试环境
• 管理工具：使用Kubernetes Cluster Federation和 Rancher 进行统一管理
• 网络配置：使用专线和VPN实现集群间的网络连通
• 应用部署：使用Helm和GitOps实现跨集群的应用部署
• 监控体系：使用Prometheus和Grafana实现跨集群的监控

4.2 混合云多集群部署案例

# 配置混合云集群
$ kubectl config use-context private-cloud

# 部署应用到私有云集群
$ kubectl apply -f – << 'EOF' apiVersion: apps/v1 kind: Deployment metadata: name: private-webapp namespace: fgedu-app spec: replicas: 3 selector: matchLabels: app: webapp environment: private template: metadata: labels: app: webapp environment: private spec: containers: - name: webapp image: harbor.fgedu.net.cn/library/webapp:v1.0.0 ports: - containerPort: 80 resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" EOF # 部署应用到公有云集群 $ kubectl config use-context public-cloud $ kubectl apply -f - << 'EOF' apiVersion: apps/v1 kind: Deployment metadata: name: public-webapp namespace: fgedu-app spec: replicas: 2 selector: matchLabels: app: webapp environment: public template: metadata: labels: app: webapp environment: public spec: containers: - name: webapp image: harbor.fgedu.net.cn/library/webapp:v1.0.0 ports: - containerPort: 80 resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" EOF # 配置混合云负载均衡 $ kubectl apply -f - << 'EOF' apiVersion: apps/v1 kind: Deployment metadata: name: hybrid-cloud-lb namespace: kube-system spec: replicas: 2 selector: matchLabels: app: hybrid-cloud-lb template: metadata: labels: app: hybrid-cloud-lb spec: containers: - name: hybrid-cloud-lb image: harbor.fgedu.net.cn/library/hybrid-cloud-lb:v1.0.0 ports: - containerPort: 80 env: - name: PRIVATE_CLOUD_URL value: "http://private-webapp.fgedu-app.svc.cluster.local" - name: PUBLIC_CLOUD_URL value: "http://public-webapp.fgedu-app.svc.cluster.local" resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" --- apiVersion: v1 kind: Service metadata: name: hybrid-cloud-lb namespace: kube-system spec: selector: app: hybrid-cloud-lb ports: - port: 80 targetPort: 80 type: LoadBalancer EOF

# 配置灾难恢复策略
$ kubectl apply -f – << 'EOF' apiVersion: types.federation.k8s.io/v1beta1 kind: FederatedDeployment metadata: name: webapp namespace: fgedu-app spec: template: metadata: labels: app: webapp spec: replicas: 5 selector: matchLabels: app: webapp template: metadata: labels: app: webapp spec: containers: - name: webapp image: harbor.fgedu.net.cn/library/webapp:v1.0.0 ports: - containerPort: 80 resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" placement: clusters: - name: primary-cluster - name: secondary-cluster overrides: - clusterName: primary-cluster clusterOverrides: - path: "/spec/replicas" value: 4 - clusterName: secondary-cluster clusterOverrides: - path: "/spec/replicas" value: 1 EOF # 模拟主集群故障 $ kubectl --context=primary-cluster cordon node primary-node1 $ kubectl --context=primary-cluster cordon node primary-node2 $ kubectl --context=primary-cluster cordon node primary-node3 # 调整副本数，将流量转移到备用集群 $ kubectl apply -f - << 'EOF' apiVersion: types.federation.k8s.io/v1beta1 kind: FederatedDeployment metadata: name: webapp namespace: fgedu-app spec: template: metadata: labels: app: webapp spec: replicas: 5 selector: matchLabels: app: webapp template: metadata: labels: app: webapp spec: containers: - name: webapp image: harbor.fgedu.net.cn/library/webapp:v1.0.0 ports: - containerPort: 80 resources: requests: cpu: "100m" memory: "256Mi" limits: cpu: "500m" memory: "512Mi" placement: clusters: - name: primary-cluster - name: secondary-cluster overrides: - clusterName: primary-cluster clusterOverrides: - path: "/spec/replicas" value: 0 - clusterName: secondary-cluster clusterOverrides: - path: "/spec/replicas" value: 5 EOF # 验证备用集群的部署 $ kubectl --context=secondary-cluster get deployments -n fgedu-app NAME READY UP-TO-DATE AVAILABLE AGE webapp 5/5 5 5 10m

#!/bin/bash
# multi_cluster_manager.sh
# from:www.itpux.com.qq113257174.wx:itpux-com
# web: `http://www.fgedu.net.cn`

# 列出所有集群
list_clusters() {
echo “=== 列出所有集群 ===”
kubectl config get-contexts
}

# 切换集群上下文
switch_cluster() {
local cluster=$1
echo “=== 切换到集群: $cluster ===”
kubectl config use-context $cluster
}

# 部署应用到指定集群
deploy_app() {
local cluster=$1
local app=$2
local namespace=$3

echo “=== 部署应用 $app 到集群 $cluster ===”
kubectl –context=$cluster apply -f $app -n $namespace
}

# 查看跨集群应用状态
check_fgapp_status() {
local namespace=$1
local app=$2

echo “=== 查看应用 $app 在所有集群中的状态 ===”
for cluster in $(kubectl config get-contexts -o name); do
echo “\n集群: $cluster”
kubectl –context=$cluster get pods -n $namespace -l app=$app
done
}

# 跨集群执行命令
exec_across_clusters() {
local command=$1

echo “=== 在所有集群中执行命令: $command ===”
for cluster in $(kubectl config get-contexts -o name); do
echo “\n集群: $cluster”
kubectl –context=$cluster $command
done
}

# 主菜单
main() {
echo “多集群管理脚本”
echo “1. 列出所有集群”
echo “2. 切换集群上下文”
echo “3. 部署应用到指定集群”
echo “4. 查看跨集群应用状态”
echo “5. 跨集群执行命令”
echo “6. 退出”
read -p “请选择操作: ” choice

case $choice in
1)
list_clusters
;;
2)
read -p “请输入集群名称: ” cluster
switch_cluster $cluster
;;
3)
read -p “请输入集群名称: ” cluster
read -p “请输入应用配置文件路径: ” app
read -p “请输入命名空间: ” namespace
deploy_app $cluster $app $namespace
;;
4)
read -p “请输入命名空间: ” namespace
read -p “请输入应用名称: ” app
check_fgapp_status $namespace $app
;;
5)
read -p “请输入要执行的命令: ” command
exec_across_clusters “$command”
;;
6)
exit 0
;;
*)
echo “无效选择”
;;
esac

main
}

main

from Linux:www.itpux.com。

Part05-风哥经验总结与分享

5.1 多集群管理最佳实践

• 统一管理平台：使用Rancher、OpenShift等工具实现多集群的统一管理
• 自动化部署：使用GitOps和CI/CD工具实现跨集群的自动化部署
• 网络规划：确保集群间的网络连通性，使用服务网格实现跨集群的服务通信
• 监控与告警：建立统一的监控体系，实现跨集群的监控和告警
• 灾备策略：制定完善的灾难恢复策略，确保在集群故障时能够快速切换

5.2 常见问题与解决方案

• 集群间网络不通：检查网络配置，确保VPN或专线连接正常
• 服务发现失败：配置正确的DNS服务，确保跨集群的服务能够被发现
• 负载均衡不均：调整负载均衡策略，确保流量在集群之间合理分配
• 数据同步延迟：使用合适的数据同步工具，确保跨集群的数据一致性
• 管理复杂度高：使用自动化工具，简化多集群的管理和操作

5.3 性能优化建议

• 网络优化：使用高速网络连接，减少集群间的网络延迟
• 资源分配：根据应用的需求，合理分配集群资源
• 负载均衡：使用智能负载均衡算法，提高资源利用率
• 缓存策略：使用缓存减少跨集群的数据传输
• 自动化扩缩容：根据负载自动调整集群资源，提高资源利用率

5.4 未来发展趋势

• 云原生多集群管理：随着云原生技术的发展，多集群管理将更加标准化和自动化
• 边缘计算集成：将边缘集群纳入多集群管理体系，实现边缘-云协同
• AI驱动的管理：使用AI技术实现多集群的智能管理和优化
• 服务网格普及：服务网格将成为跨集群服务通信的标准解决方案
• 标准化和互操作性：多集群管理的标准将逐渐形成，提高不同平台的互操作性

风学习交流加群风哥QQ113257174哥提示：多集群管理是未来企业级Kubernetes部署的重要方向，需要持续学习和实践，以适应不断变化的业务需求和技术环境。