简介：本文侧重介绍了通过 ACK One 的多集群应用分发功能，可以帮助企业管理多集群环境，通过多集群主控示例提供的统一的应用下发入口，实现应用的多集群分发，差异化配置，工作流管理等分发策略。结合 GTM 全局流量管理，快速搭建管理两地三中心的应用容灾系统。
作者：宇汇，壮怀，先河
概述
两地三中心是指在两个城市部署三个业务处理中心，即：生产中心、同城容灾中心、异地容灾中心。在一个城市部署 2 套环境形成同城双中心，同时处理业务并通过高速链路实现数据同步，可切换运行。在另一城市部署1套环境做异地灾备中心，做数据备份，当双中心同时故障时，异地灾备中心可切换处理业务。两地三中心容灾方案可以极大程度的保证业务的连续运行。
使用 ACK One 的多集群管理应用分发功能，可以帮助企业统一管理 3 个 K8s 集群，实现应用在 3 个 K8s 集群快速部署升级，同时实现应用在 3 个 K8s 集群上的差异化配置。配合使用 GTM（全局流量管理）可以实现在故障发生时业务流量在 3 个 K8s 集群的自动切换。对 RDS 数据层面的数据复制，本实践不做具体介绍，可参考 DTS 数据传输服务。
方案架构
前提条件
开启多集群管理主控实例[1]
通过管理关联集群[2]，添加 3 个 K8s 集群到主控实例中，构建两地三中心。本实践中，作为示例，在北京部署 2 个 K8s 集群（cluster1-beijing 和 cluster2-beijing），在杭州部署 1 个 K8s 集群（cluster1-hangzhou）。
创建 GTM 实例[3]
应用部署
通过 ACK One 主控实例的应用分发功能[4]，在 3 个 K8s 集群中分发应用。对比传统的脚本部署，使用 ACK One 的应用分发可获得如下收益。
本实践中，示例应用为 web 应用，包含 K8s Deployment/Service/Ingress/Configmap 资源，Service/Ingress 对外暴露服务，Deployment 读取 Configmap 中的配置参数。通过创建应用分发规则，将应用分发到 3 个 K8s 集群，包括 2 个北京集群，1 个杭州集群，实现两地三中心。分发过程中对 deployment 和 configmap 资源做差异化配置，以适应不用地点的集群，同时分发过程实现人工审核的灰度控制，限制错误的爆炸半径。
1. 执行一下命令创建命名空间 demo。
kubectl create namespace demo
2. 使用以下内容，创建 app-meta.yaml 文件。
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }3. 执行以下命令，在主控实例上部署应用 web-demo。注意：在主控实例上创建 kube 资源并不会下发到子集群，此 kube 资源作为原数据，被后续 Application（步骤 4b）中引用。
kubectl apply -f app-meta.yaml
4. 创建应用分发规则。
a. 执行以下命令，查看主控实例管理的关联集群，确定应用的分发目标
kubectl amc get managedcluster
预期输出：
Name Alias HubAcceptedmanagedcluster-cxxx cluster1-hangzhou truemanagedcluster-cxxx cluster2-beijing truemanagedcluster-cxxx cluster1-beijing true
b. 使用以下内容，创建应用分发规则 app.yaml。替换示例中的和 managedcluster-cxxx 为实际待发布集群名称。分发规则定义的最佳实践在注释中说明。
在 app.yaml 中，包含以下资源类型：Policy (type:topology) 分发目标，Policy (type: override)差异化规则， Workflow 工作流，Application 应用。具体可参考：应用复制分发[5]、应用分发差异化配置[6]和应用集群间灰度分发[7]。
apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: cluster1-beijing namespace: demotype: topologyproperties: clusters: ["<managedcluster-cxxx>"] #分发目标集群1 cluster1-beijing---apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: cluster2-beijing namespace: demotype: topologyproperties: clusters: ["<managedcluster-cxxx>"] #分发目标集群2 cluster2-beijing---apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: cluster1-hangzhou namespace: demotype: topologyproperties: clusters: ["<managedcluster-cxxx>"] #分发目标集群3 cluster1-hangzhou---apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: override-env-cluster2-beijing namespace: demotype: overrideproperties: components: - name: "deployment" traits: - type: env properties: containerName: web-demo env: ENV_NAME: cluster2-beijing #对集群cluster2-beijing的deployment做环境变量的差异化配置---apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: override-env-cluster1-hangzhou namespace: demotype: overrideproperties: components: - name: "deployment" traits: - type: env properties: containerName: web-demo env: ENV_NAME: cluster1-hangzhou #对集群cluster1-hangzhou的deployment做环境变量的差异化配置---apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: override-replic-cluster1-hangzhou namespace: demotype: overrideproperties: components: - name: "deployment" traits: - type: scaler properties: replicas: 1 #对集群cluster1-hangzhou的deployment做副本数的差异化配置---apiVersion: core.oam.dev/v1alpha1kind: Policymetadata: name: override-configmap-cluster1-hangzhou namespace: demotype: overrideproperties: components: - name: "configmap" traits: - type: json-merge-patch #对集群cluster1-hangzhou的deployment做configmap的差异化配置 properties: data: config.json: | { database-address: "hangzhou-db.pg.aliyun.com" }---apiVersion: core.oam.dev/v1alpha1kind: Workflowmetadata: name: deploy-demo namespace: demosteps: #顺序部署cluster1-beijing，cluster2-beijing，cluster1-hangzhou。 - type: deploy name: deploy-cluster1-beijing properties: policies: ["cluster1-beijing"] - type: deploy name: deploy-cluster2-beijing properties: auto: false #部署cluster2-beijing前需要人工审核 policies: ["override-env-cluster2-beijing", "cluster2-beijing"] #在部署cluster2-beijing时做环境变量的差异化 - type: deploy name: deploy-cluster1-hangzhou properties: policies: ["override-env-cluster1-hangzhou", "override-replic-cluster1-hangzhou", "override-configmap-cluster1-hangzhou", "cluster1-hangzhou"] #在部署cluster2-beijing时做环境变量，副本数，configmap的差异化---apiVersion: core.oam.dev/v1beta1kind: Applicationmetadata: annotations: app.oam.dev/publishVersion: version8 name: web-demo namespace: demospec: components: - name: deployment #独立引用deployment，方便差异化配置 type: ref-objects properties: objects: - apiVersion: apps/v1 kind: Deployment name: web-demo - name: configmap #独立引用configmap，方便差异化配置 type: ref-objects properties: objects: - apiVersion: v1 kind: ConfigMap name: web-demo - name: same-resource #不做差异化配置 type: ref-objects properties: objects: - apiVersion: v1 kind: Service name: web-demo - apiVersion: networking.k8s.io/v1 kind: Ingress name: web-demo workflow: ref: deploy-demo5. 执行以下命令，在主控实例上部署分发规则 app.yaml。
kubectl apply -f app.yaml
6. 查看应用的部署状态。
kubectl get app web-demo -n demo
预期输出，workflowSuspending 表示部署暂停
NAME COMPONENT TYPE PHASE HEALTHY STATUS AGEweb-demo deployment ref-objects workflowSuspending true 47h
7. 查看应用在各个集群上的运行状态
kubectl amc get deployment web-demo -n demo -m all
预期输出：
Run on ManagedCluster managedcluster-cxxx (cluster1-hangzhou)No resources found in demo namespace #第一次新部署应用，工作流还没有开始部署cluster1-hangzhouRun on ManagedCluster managedcluster-cxxx (cluster2-beijing)No resources found in demo namespace #第一次新部署应用，工作流还没有开始部署cluster2-beijiing，等待人工审核Run on ManagedCluster managedcluster-cxxx (cluster1-beijing)NAME READY UP-TO-DATE AVAILABLE AGEweb-demo 5/5 5 5 47h #Deployment在cluster1-beijing集群上运行正常
8. 人工审核通过，部署集群 cluster2-beijing，cluster1-hangzhou。
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }09. 查看应用的部署状态。
kubectl get app web-demo -n demo
预期输出，running 表示应用运行正常
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }210. 查看应用在各个集群上的运行状态
kubectl amc get deployment web-demo -n demo -m all
预期输出：
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }411. 查看应用在各个集群上的 Ingress 状态
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }5预期结果，每个集群的 Ingress 运行正常，公网 IP 分配成功。
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }6流量管理
通过配置全局流量管理，自动检测应用运行状态，并在异常发生时，自动切换流量到监控集群。
1. 配置全局流量管理实例，web-demo.example.com 为示例应用的域名，请替换为实际应用的域名，并设置 DNS 解析到全局流量管理的 CNAME 接入域名。
2. 在已创建的 GTM 示例中，创建 2 个地址池：
a、pool-beijing：包含 2 个北京集群的 Ingress IP 地址，负载均衡策略为返回全部地址，实现北京 2 个集群的负载均衡。Ingress IP 地址可通过在主控实例上运行 “kubectl amc get ingress -n demo -m all” 获取。
b、pool-hangzhou：包含 1 个杭州集群的 Ingress IP 地址。
3. 在地址池中开启健康检查，检查失败的地址将从地址池中移除，不再接收流量。
4. 配置访问策略，设置主地址池为北京地址池，备地址池为杭州地址池。正常流量都有北京集群应用处理，当所有北京集群应用不可用时，自动切换到杭州集群应用处理。
部署验证
1. 正常情况，所有有流量都有北京的 2 个集群上的应用处理，每个集群各处理 50% 流量。
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }72. 当集群 cluster1-beijing 上的应用异常时，GTM 将所有的流量路由到 cluster2-bejing 集群处理。
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }83. 当集群 cluster1-beijing 和 cluster2-beijing 上的应用同时异常时，GTM 将流量路由到 cluster1-hangzhou 集群处理。
apiVersion: apps/v1kind: Deploymentmetadata: labels: app: web-demo name: web-demo namespace: demospec: replicas: 5 selector: matchLabels: app: web-demo template: metadata: labels: app: web-demo spec: containers: - image: acr-multiple-clusters-registry.cn-hangzhou.cr.aliyuncs.com/ack-multiple-clusters/web-demo:0.4.0 name: web-demo env: - name: ENV_NAME value: cluster1-beijing volumeMounts: - name: config-file mountPath: "/config-file" readOnly: true volumes: - name: config-file configMap: items: - key: config.json path: config.json name: web-demo---apiVersion: v1kind: Servicemetadata: name: web-demo namespace: demo labels: app: web-demospec: selector: app: web-demo ports: - protocol: TCP port: 80 targetPort: 8080---apiVersion: networking.k8s.io/v1kind: Ingressmetadata: name: web-demo namespace: demo labels: app: web-demospec: rules: - host: web-demo.example.com http: paths: - path: / pathType: Prefix backend: service: name: web-demo port: number: 80---apiVersion: v1kind: ConfigMapmetadata: name: web-demo namespace: demo labels: app: web-demodata: config.json: | { database-host: "beijing-db.pg.aliyun.com" }9总结
本文侧重介绍了通过 ACK One 的多集群应用分发功能，可以帮助企业管理多集群环境，通过多集群主控示例提供的统一的应用下发入口，实现应用的多集群分发，差异化配置，工作流管理等分发策略。结合 GTM 全局流量管理，快速搭建管理两地三中心的应用容灾系统。
除多集群应用分发外，ACK One 更是支持连接并管理任何地域、任何基础设施上的 Kubernetes 集群，提供一致的管理和社区兼容的 API，支持对计算、网络、存储、安全、监控、日志、作业、应用、流量等进行统一运维管控。阿里云分布式云容器平台（简称 ACK One）是面向混合云、多集群、分布式计算、容灾等场景推出的企业级云原生平台。更多内容可以查看产品介绍分布式云容器平台 ACK One[8]。
相关链接
[1] 开启多集群管理主控实例：
https://help.aliyun.com/document\_detail/384048.html
[2] 通过管理关联集群：
https://help.aliyun.com/document\_detail/415167.html
[3] 创建 GTM 实例：
https://dns.console.aliyun.com/#/gtm2/list
[4] 应用分发功能：
https://help.aliyun.com/document\_detail/412652.html
[5] 应用复制分发：
https://help.aliyun.com/document\_detail/412653.html
[6] 应用分发差异化配置：
https://help.aliyun.com/document\_detail/412707.html
[7] 应用集群间灰度分发：
https://help.aliyun.com/document\_detail/412787.html
[8] 分布式云容器平台 ACK One：
https://www.aliyun.com/product/aliware/adcp
原文链接
本文为阿里云原创内容，未经允许不得转载。
原文：https://juejin.cn/post/7096766204634906637