阅读量:0
文章目录
安装方式一：
1. 快速部署一个简单的 EMQX 集群：
添加 helm 仓库：
$ helm repo add emqx https://repos.emqx.io/charts # 查询 EMQX $ helm search repo emqx 启动 EMQX 集群，设置 service.type=NodePort
$ helm install my-emqx emqx/emqx --set service.type=NodePort NAME: my-emqx LAST DEPLOYED: Thu Jul 11 10:56:16 2024 NAMESPACE: default STATUS: deployed REVISION: 1 TEST SUITE: None 查看 EMQX 集群情况：
$ kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES my-emqx-0 0/1 ContainerCreating 0 12s <none> master-01 <none> <none> my-emqx-1 0/1 ContainerCreating 0 12s <none> worker-02 <none> <none> my-emqx-2 0/1 ContainerCreating 0 12s <none> worker-01 <none> <none> $ kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES my-emqx-0 1/1 Running 0 3m7s 172.20.32.158 master-01 <none> <none> my-emqx-1 1/1 Running 0 3m7s 172.20.58.232 worker-02 <none> <none> my-emqx-2 1/1 Running 0 3m7s 172.20.85.250 worker-01 <none> <none> $ kubectl exec -it my-emqx-0 -- emqx_ctl status Node 'my-emqx@my-emqx-0.my-emqx-headless.default.svc.cluster.local' 5.7.1 is started $ kubectl exec -it my-emqx-0 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['my-emqx@my-emqx-0.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-1.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-2.my-emqx-headless.default.svc.cluster.local'], stopped_nodes => []} 查看 EMQX service：
$ kubectl get svc -o wide NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) service/my-emqx NodePort 10.68.183.22 <none> 1883:32596/TCP,8883:31933/TCP,8083:32522/TCP,8084:32535/TCP,18083:32717/TCP 10m app.kubernetes.io/instance=my-emqx,app.kubernetes.io/name=emqx service/my-emqx-headless ClusterIP None <none> 1883/TCP,8883/TCP,8083/TCP,8084/TCP,18083/TCP,4370/TCP 10m app.kubernetes.io/instance=my-emqx,app.kubernetes.io/name=emqx # 也可以执行该命令查看全部 $ sudo kubectl get all -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES pod/my-emqx-0 1/1 Running 0 10m 172.20.32.158 master-01 <none> <none> pod/my-emqx-1 1/1 Running 0 10m 172.20.58.232 worker-02 <none> <none> pod/my-emqx-2 1/1 Running 0 10m 172.20.85.250 worker-01 <none> <none> NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) service/my-emqx NodePort 10.68.183.22 <none> 1883:32596/TCP,8883:31933/TCP,8083:32522/TCP,8084:32535/TCP,18083:32717/TCP 10m app.kubernetes.io/instance=my-emqx,app.kubernetes.io/name=emqx service/my-emqx-headless ClusterIP None <none> 1883/TCP,8883/TCP,8083/TCP,8084/TCP,18083/TCP,4370/TCP 10m app.kubernetes.io/instance=my-emqx,app.kubernetes.io/name=emqx NAME READY AGE CONTAINERS IMAGES statefulset.apps/my-emqx 3/3 10m emqx emqx/emqx:5.7.1 可以看到 my-emqx 的 18083 端口对应的宿主机的 32717 端口。（NodePort 在每次部署的时候都会变化，以实际部署时为准。）
访问 Kubernetes 的任意一台节点 IP 的 32717 端口，输入默认用户名：admin，默认密码：public，登陆 EMQX dashboard。
删除 EMQX 集群：
$ helm uninstall my-emqx release "my-emqx" uninstalled 2. 部署一个持久化的 EMQX 集群：
EMQX 通过 创建 PVC 资源挂载 /opt/emqx/data/mnesia 目录实现持久化 pods，在部署 EMQX 之前，用户需要部署 Haproxy 或 Nginx-PLUS 等负载均衡器，并自行在 Kubernetes 中创建 PVC 资源或是 Storage Classes 资源
如果用户部署了 PVC 资源，那么设置 persistence.existingClaim=your_pv_name
$ helm install my-emqx emqx/emqx --set persistence.enabled=true --set persistence.existingClaim=your_pv_name 如果用户部署了 Storage Classes 资源，那么设置 persistence.storageClass=your_storageClass_name
$ helm install my-emqx emqx/emqx --set persistence.enabled=true --set persistence.storageClass=your_storageClass_name 查看 EMQX 集群情况
$ kubectl get pods NAME READY STATUS RESTARTS AGE my-emqx-0 1/1 Running 0 56 smy-emqx-1 1/1 Running 0 40s my-emqx-2 1/1 Running 0 21s $ kubectl exec -it my-emqx-0 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['my-emqx@my-emqx-0.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-1.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-2.my-emqx-headless.default.svc.cluster.local'], stopped_nodes => []} 以 Storage Classes 为例，可以看到 PVC 资源已经成功的建立
$ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-data-my-emqx-0 Bound pvc-8094cd75-adb5-11e9-80cc-0697b59e8064 1Gi RWO gp2 2m11s emqx-data-my-emqx-1 Bound pvc-9325441d-adb5-11e9-80cc-0697b59e8064 1Gi RWO gp2 99s emqx-data-my-emqx-2 Bound pvc-ad425e9d-adb5-11e9-80cc-0697b59e8064 1Gi RWO gp2 56s 集群会将 EMQX 的 /opt/emqx/data/mnesia 目录挂载到 PVC 中，当 Pods 被重新调度之后，EMQX 会从 /opt/emqx/data/mnesia 目录中获取数据并恢复。查看 EMQX 的 ClusterIP
$ kubectl get svc NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE my-emqx ClusterIP 10.100.205.13 <none> 1883/TCP,8883/TCP,8081/TCP,8083/TCP,8084/TCP,18083/TCP 26m my-emqx-headless ClusterIP None <none> 1883/TCP,8883/TCP,8081/TCP,8083/TCP,8084/TCP,18083/TCP 26m 可以看到 my-emqx 的 ClusterIP 为 10.100.205.13 （ClusterIP 在每次部署的时候都会变化，以实际部署时为准。）
将负载均衡监听的 URL 的 1883、8883、8081、8083、8084、18083 端口转发到 my-emqx 的 ClusterIP，如果有 TLS 连接的需要，推荐在负载均衡器终结 SSL 连接。客户端与负载均衡器之间 TLS 安全连接，LB 与 EMQX 之间普通 TCP 连接。
访问 URL:18083，输入默认用户名：admin，默认密码：public，登陆 EMQX dashboard。
使用 helm upgrade 命令可以轻松扩展 EMQX 集群，下面以增加 EMQX 节点为例展示 helm upgrade 命令
# 将 EMQX 的节点数量变更为5个 # 注意：EMQX 的节点数量建议为单数 $ helm upgrade --set replicaCount=5 my-emqx emqx/emqx Release "my-emqx" has been upgraded. Happy Helming! $ kubectl get pods NAME READY STATUS RESTARTS AGE my-emqx-0 1/1 Running 0 4m25s my-emqx-1 1/1 Running 0 4m14s my-emqx-2 1/1 Running 0 4m my-emqx-3 1/1 Running 0 31s my-emqx-4 1/1 Running 0 15s $ kubectl exec -it my-emqx-0 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['my-emqx@my-emqx-0.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-1.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-2.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-3.my-emqx-headless.default.svc.cluster.local', 'my-emqx@my-emqx-4.my-emqx-headless.default.svc.cluster.local'], stopped_nodes => []} 删除 EMQX 集群
$ helm uninstall my-emqx release "my-emqx" uninstalled 注意：EMQX 集群删除掉之后 PVC 资源不会自动释放掉，以便恢复 EMQX，确认不需要恢复后需要手动删除 PVC 资源
$ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-data-my-emqx-0 Bound pvc-8094cd75-adb5-11e9-80cc-0697b59e8064 1Gi RWO gp2 84m emqx-data-my-emqx-1 Bound pvc-9325441d-adb5-11e9-80cc-0697b59e8064 1Gi RWO gp2 84m emqx-data-my-emqx-2 Bound pvc-ad425e9d-adb5-11e9-80cc-0697b59e8064 1Gi RWO gp2 83m emqx-data-my-emqx-3 Bound pvc-b6c5a565-adbd-11e9-80cc-0697b59e8064 1Gi RWO gp2 25m emqx-data-my-emqx-4 Bound pvc-c626cafd-adbd-11e9-80cc-0697b59e8064 1Gi RWO gp2 25m $ kubectl delete pvc emqx-data-my-emqx-0 emqx-data-my-emqx-1 emqx-data-my-emqx-2 emqx-data-my-emqx-3 emqx-data-my-emqx-4 persistentvolumeclaim "emqx-data-my-emqx-0" deleted persistentvolumeclaim "emqx-data-my-emqx-1" deleted persistentvolumeclaim "emqx-data-my-emqx-2" deleted persistentvolumeclaim "emqx-data-my-emqx-3" deleted persistentvolumeclaim "emqx-data-my-emqx-4" deleted 实测可以执行的命令：
helm install my-emqx emqx/emqx -n emqx --create-namespace --set service.type=NodePort --set service.nodePorts.dashboard=32717 --set service.nodePorts.mqtt=32596 --set persistence.storageClass=nfs 3. 部署 EMQX Edge 集群和 EMQX 企业版集群：
EMQX Edge：部署 EMQX Edge 集群指定 image.repository=emqx/emqx-edge，其他设置与部署 EMQX 集群保持一致
$ helm install my-emqx-edge emqx/emqx --set image.repository=emqx/emqx$ kubectl get pods NAME READY STATUS RESTARTS AGEmy-emqx-edge-0 1/1 Running 0 35smy-emqx-edge-1 1/1 Running 0 23smy-emqx-edge-2 1/1 Running 0 9s EMQX EE：部署 EMQX 企业版 集群首先需要前往 www.emqx.com 申请并下载 License 文件，并将 License 文件创建为 Secret 资源
$ kubectl create secret generic your-license-secret-name --from-file=/path/to/emqx.lic 然后在部署时指定 repo 为 emqx/emqx-ee, emqxLicneseSecretName=your-license-secret-name, 其他设置与部署 EMQX 集群保持一致
$ helm install my-emqx-ee emqx/emqx-ee emqxLicneseSecretName=your-license-secret-name 安装方式二：定制化部署
参考自：从零开始建立 EMQX MQTT 服务器的 K8S 集群
1. 使用 Pod 直接部署 EMQX Broker
EMQX Broker 在 docker hub 上提供了镜像, 因此可以很方便的在单个的 pod 上部署 EMQX Broker，使用 kubectl run 命令创建一个运行着 EMQX Broker 的 Pod：
$ kubectl run emqx --image=emqx/emqx:v4.1-rc.1 --generator=run-pod/v1 error: unknown flag: --generator See 'kubectl run --help' for usage. $ kubectl run emqx --image=emqx/emqx:v4.1-rc.1 pod/emqx created 查看 EMQX Broker 的状态：
$ kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES emqx 0/1 ContainerCreating 0 8s <none> master-01 <none> <none> $ kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES emqx 1/1 Running 0 97s 172.20.32.139 master-01 <none> <none> $ kubectl exec emqx -- emqx_ctl status Node 'emqx@172.20.32.139' is started emqx 4.1-rc.1 is running 删除 Pod：
$ kubectl delete pods emqx pod "emqx" deleted Pod 并不是被设计成一个持久化的资源，它不会在调度失败，节点崩溃，或者其他回收中（比如因为资源的缺乏，或者其他的维护中）幸存下来，因此，还需要一个控制器来管理 Pod。
2. 使用 Deoloyment 部署 Pod
Deployment 为 Pod 和 ReplicaSet 提供了一个声明式定义（declarative）方法，用来替代以前的 ReplicationController 来方便的管理应用。典型的应用场景包括：
- 定义Deployment来创建Pod和ReplicaSet
- 滚动升级和回滚应用
- 扩容和缩容
- 暂停和继续Deployment
使用 Deployment 部署一个 EMQX Broker Pod：
$ vim deployment.yaml apiVersion: apps/v1 kind: Deployment metadata: name: emqx-deployment labels: app: emqx spec: replicas: 1 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 部署 Deployment：
$ kubectl apply -f deployment.yaml deployment.apps/emqx-deployment created 查看部署情况：
$ kubectl get deployment NAME READY UP-TO-DATE AVAILABLE AGE emqx-deployment 0/1 1 0 49s $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-75bd4f75b6-wf9sh 1/1 Running 0 73s $ kubectl exec pod/emqx-deployment-75bd4f75b6-wf9sh -- emqx_ctl status Node 'emqx-deployment-75bd4f75b6-wf9sh@172.20.58.199' is started emqx 4.1-rc.1 is running $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-75bd4f75b6-8njhv 1/1 Running 0 55s 尝试手动删除 Pod：
$ kubectl delete pods emqx-deployment-75bd4f75b6-8njhv pod "emqx-deployment-75bd4f75b6-8njhv" deleted $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-68fcb4bfd6-2nhh6 1/1 Running 0 59s 输出结果表明成功用 Deployment 部署了 EMQX Broker Pod，即使是此 Pod 被意外终止，Deployment 也会重新创建一个新的 Pod。
3. 使用 Services 公开 EMQX Broker Pod 服务
Kubernetes Pods 是有生命周期的。他们可以被创建，而且销毁不会再启动。 如果使用 Deployment 来运行应用程序，则它可以动态创建和销毁 Pod。
每个 Pod 都有自己的 IP 地址，但是在 Deployment 中，在同一时刻运行的 Pod 集合可能与稍后运行该应用程序的 Pod 集合不同。
这导致了一个问题：如果使用 EMQX Broker Pod 为 MQTT 客户端提供服务，那么客户端应该如何如何找出并跟踪要连接的 IP 地址，以便客户端使用 EMQX Broker 服务呢？
答案是：Service
Service 是将运行在一组 Pods 上的应用程序公开为网络服务的抽象方法。
使用 Service 将 EMQX Broker Pod 公开为网络服务：
vim service.yaml apiVersion: v1 kind: Service metadata: name: emqx-service spec: selector: app: emqx ports: - name: mqtt port: 1883 protocol: TCP targetPort: mqtt - name: mqttssl port: 8883 protocol: TCP targetPort: mqttssl - name: mgmt port: 8081 protocol: TCP targetPort: mgmt - name: ws port: 8083 protocol: TCP targetPort: ws - name: wss port: 8084 protocol: TCP targetPort: wss - name: dashboard port: 18083 protocol: TCP targetPort: dashboard 部署 Service：
$ kubectl apply -f service.yaml service/emqx-service created 查看部署情况
$ kubectl get svc NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE emqx-service ClusterIP 10.68.228.164 <none> 1883/TCP,8883/TCP,8081/TCP,8083/TCP,8084/TCP,18083/TCP 16s 使用 Service 提供的 IP 查看 EMQX Broker 的 API
$ curl 10.68.228.164:8081/status Node emqx-deployment-75bd4f75b6-8njhv@172.20.32.140 is started emqx is running 至此，单个 EMQX Broker 节点在 kubernetes 上部署完毕，通过 Deployment 管理 EMQX Broker Pod，通过 Service 将 EMQX Broker 服务暴露出去。
4. 通过 kubernetes 自动集群 EMQX MQTT 服务器
上文中通过 Deployment 部署了单个的 EMQX Broker Pod，通过 Deployment 扩展 Pod 的数量是极为方便的，执行 kubectl scale deployment ${deployment_name} --replicas ${numer} 命令即可扩展 Pod 的数量，下面将 EMQX Broker Pod 扩展为 3 个：
$ kubectl scale deployment emqx-deployment --replicas 3 deployment.apps/emqx-deployment scaled $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-75bd4f75b6-8njhv 1/1 Running 0 8m41s emqx-deployment-75bd4f75b6-kjv22 0/1 ContainerCreating 0 12s emqx-deployment-75bd4f75b6-thcbv 1/1 Running 0 12s $ kubectl exec emqx-deployment-75bd4f75b6-8njhv -- emqx_ctl status Node 'emqx-deployment-75bd4f75b6-8njhv@172.20.32.140' is started emqx 4.1-rc.1 is running $ kubectl exec emqx-deployment-75bd4f75b6-8njhv -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx-deployment-75bd4f75b6-8njhv@172.20.32.140'], stopped_nodes => []} 可以看到 EMQX Broker Pod 的数量被扩展为 3 个，但是每个 Pod 都是独立的，并没有集群，接下来尝试通过 kubernetes 自动集群 EMQX Broker Pod。
5. 修改 EMQX Broker 的配置
查看 EMQX Broker 文档中关于 自动集群 的内容，可以看到需要修改 EMQX Broker 的配置：
cluster.discovery = kubernetes cluster.kubernetes.apiserver = http://10.110.111.204:8080 cluster.kubernetes.service_name = ekka cluster.kubernetes.address_type = ip cluster.kubernetes.app_name = ekka 其中 cluster.kubernetes.apiserver 为 kubernetes apiserver 的地址，可以通过 kubectl cluster-info 命令获取，cluster.kubernetes.service_name 为上文中 Service 的 name， cluster.kubernetes.app_name 为 EMQX Broker 的 node.name 中 @ 符号之前的部分，所以还需要将集群中 EMQX Broker 设置为统一的 node.name 的前缀。
EMQX Broker 的 docker 镜像提供了通过环境变量修改配置的功能，具体可以查看 docker hub 或 Github。
$ vim deployment.yaml apiVersion: apps/v1 kind: Deployment metadata: name: emqx-deployment labels: app: emqx spec: replicas: 3 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 env: - name: EMQX_NAME value: emqx - name: EMQX_CLUSTER__DISCOVERY value: k8s - name: EMQX_CLUSTER__K8S__APP_NAME value: emqx - name: EMQX_CLUSTER__K8S__SERVICE_NAME value: emqx-service - name: EMQX_CLUSTER__K8S__APISERVER value: "https://kubernetes.default.svc:443" - name: EMQX_CLUSTER__K8S__NAMESPACE value: default 因为 kubectl scale deployment ${deployment_name} --replicas ${numer} 命令不会修改 yaml 文件，所以修改 yaml 时需要设置 spec.replicas: 3 。
Pod 中内建 kubernetes 的 DNS 规则，所以 https://kubernetes.default.svc:443 会被解析为 kubernetes apiserver 的地址。
删除之前的 Deployment，重新部署：
$ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted $ root@k8s-master01:/home/ubuntu/huiq/emqx# kubectl apply -f deployment.yaml deployment.apps/emqx-deployment created 6. 赋予 Pod 访问 kubernetes apiserver 的权限
上文部署 Deployment 之后，查看 EMQX Broker 的状态，可以看到 EMQX Broker 虽然成功启动了，但是依然没有集群成功，查看 EMQX Broker Pod 的 log：
$ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-64d5cc575c-dsrgj 1/1 Running 0 41s emqx-deployment-64d5cc575c-k4twn 1/1 Running 0 41s emqx-deployment-64d5cc575c-ktdqk 1/1 Running 0 41s $ kubectl exec emqx-deployment-64d5cc575c-dsrgj -- emqx_ctl status Node 'emqx@172.20.58.200' is started emqx 4.1-rc.1 is running $ kubectl exec emqx-deployment-64d5cc575c-dsrgj -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@172.20.58.200'],stopped_nodes => []} root@k8s-master01:/home/ubuntu/huiq/emqx# kubectl logs emqx-deployment-64d5cc575c-dsrgj ... (emqx@172.20.58.200)1> 2024-07-11 08:14:37.525 [error] Ekka(AutoCluster): Discovery error: {403,"{\"kind\":\"Status\",\"apiVersion\":\"v1\",\"metadata\":{},\"status\":\"Failure\",\"message\":\"endpoints \\\"emqx-service\\\" is forbidden: User \\\"system:serviceaccount:default:default\\\" cannot get resource \\\"endpoints\\\" in API group \\\"\\\" in the namespace \\\"default\\\"\",\"reason\":\"Forbidden\",\"details\":{\"name\":\"emqx-service\",\"kind\":\"endpoints\"},\"code\":403}\n"} ... Pod 因为权限问题在访问 kubernetes apiserver 的时候被拒绝，返回 HTTP 403，所以集群失败。
普通 Pod 是无法访问 kubernetes apiserver 的，解决这个问题有两种方法，一种是开放 kubernetes apiserver 的 http 接口，但是这种方法存在一定的安全隐患，另外一种是通过 ServiceAccount、Role 和 RoleBinding 配置 RBAC 鉴权。
$ vim rbac.yaml apiVersion: v1 kind: ServiceAccount metadata: namespace: default name: emqx --- kind: Role apiVersion: rbac.authorization.kubernetes.io/v1beta1 metadata: namespace: default name: emqx rules: - apiGroups: - "" resources: - endpoints verbs: - get - watch - list --- kind: RoleBinding apiVersion: rbac.authorization.kubernetes.io/v1beta1 metadata: namespace: default name: emqx subjects: - kind: ServiceAccount name: emqx namespace: default roleRef: kind: Role name: emqx apiGroup: rbac.authorization.kubernetes.io 部署相应的资源：
$ kubectl apply -f rbac.yaml serviceaccount/emqx created resource mapping not found for name: "emqx" namespace: "default" from "rbac.yaml": no matches for kind "Role" in version "rbac.authorization.kubernetes.io/v1beta1" ensure CRDs are installed first resource mapping not found for name: "emqx" namespace: "default" from "rbac.yaml": no matches for kind "RoleBinding" in version "rbac.authorization.kubernetes.io/v1beta1" ensure CRDs are installed first $ kubectl explain role GROUP: rbac.authorization.k8s.io KIND: Role VERSION: v1 $ kubectl explain RoleBinding GROUP: rbac.authorization.k8s.io KIND: RoleBinding VERSION: v1 解决上面的问题，重新编辑 rbac.yaml 文件：
apiVersion: v1 kind: ServiceAccount metadata: namespace: default name: emqx --- kind: Role apiVersion: rbac.authorization.k8s.io/v1 metadata: namespace: default name: emqx rules: - apiGroups: - "" resources: - endpoints verbs: - get - watch - list --- kind: RoleBinding apiVersion: rbac.authorization.k8s.io/v1 metadata: namespace: default name: emqx subjects: - kind: ServiceAccount name: emqx namespace: default roleRef: kind: Role name: emqx apiGroup: rbac.authorization.k8s.io $ kubectl delete serviceaccount emqx serviceaccount "emqx" deleted $ kubectl delete role emqx role.rbac.authorization.k8s.io "emqx" deleted $ kubectl delete rolebinding emqx rolebinding.rbac.authorization.k8s.io "emqx" deleted $ kubectl apply -f rbac.yaml serviceaccount/emqx created role.rbac.authorization.k8s.io/emqx created rolebinding.rbac.authorization.k8s.io/emqx created $ kubectl get serviceaccount NAME SECRETS AGE emqx 0 16s $ kubectl get role NAME CREATED AT emqx 2024-07-11T08:45:48Z $ kubectl get rolebinding NAME ROLE AGE emqx Role/emqx 51s 修改 Deployment 的 yaml 文件，增加 spec.template.spec.serviceAccountName，并重新部署：
vim deployment.yaml apiVersion: apps/v1 kind: Deployment metadata: name: emqx-deployment labels: app: emqx spec: replicas: 3 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 env: - name: EMQX_NAME value: emqx - name: EMQX_CLUSTER__DISCOVERY value: kubernetes - name: EMQX_CLUSTER__K8S__APP_NAME value: emqx - name: EMQX_CLUSTER__K8S__SERVICE_NAME value: emqx-service - name: EMQX_CLUSTER__K8S__APISERVER value: "https://kubernetes.default.svc:443" - name: EMQX_CLUSTER__K8S__NAMESPACE value: default $ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted $ kubectl apply -f deployment.yaml deployment.apps/emqx-deployment created 查看状态：
$ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-7c4675f785-6r6dg 1/1 Running 0 28s emqx-deployment-7c4675f785-cqbgh 1/1 Running 0 28s emqx-deployment-7c4675f785-nghf6 1/1 Running 0 28s $ kubectl exec emqx-deployment-7c4675f785-6r6dg -- emqx_ctl status Node 'emqx@127.0.0.1' not responding to pings. command terminated with exit code 1 尝试解决：本以为是文章里下划线写多了导致的，虽然 emqx 可以运行了但不是集群模式
$ vim deployment.yaml apiVersion: apps/v1 kind: Deployment metadata: name: emqx-deployment labels: app: emqx spec: replicas: 3 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 env: - name: EMQX_NAME value: emqx - name: EMQX_CLUSTER_DISCOVERY value: kubernetes - name: EMQX_CLUSTER_K8S_APP_NAME value: emqx - name: EMQX_CLUSTER_K8S_SERVICE_NAME value: emqx-service - name: EMQX_CLUSTER_K8S_APISERVER value: "https://kubernetes.default.svc:443" - name: EMQX_CLUSTER_K8S_NAMESPACE value: default $ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted $ kubectl apply -f deployment.yaml deployment.apps/emqx-deployment created $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-794478ff56-2rqwp 1/1 Running 0 19s emqx-deployment-794478ff56-4jnmt 1/1 Running 0 19s emqx-deployment-794478ff56-m4mk4 1/1 Running 0 19s $ kubectl exec emqx-deployment-794478ff56-2rqwp -- emqx_ctl status Node 'emqx@172.20.85.238' is started emqx 4.1-rc.1 is running $ kubectl exec emqx-deployment-794478ff56-bzck8 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@172.20.85.244'],stopped_nodes => []} 最终解决：文章里给留了个大坑，需要将 kubernetes 改为 k8s
$ vim deployment.yaml apiVersion: apps/v1 kind: Deployment metadata: name: emqx-deployment labels: app: emqx spec: replicas: 3 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 env: - name: EMQX_NAME value: emqx - name: EMQX_CLUSTER__DISCOVERY value: k8s - name: EMQX_CLUSTER__K8S__APP_NAME value: emqx - name: EMQX_CLUSTER__K8S__SERVICE_NAME value: emqx-service - name: EMQX_CLUSTER__K8S__APISERVER value: "https://kubernetes.default.svc:443" - name: EMQX_CLUSTER__K8S__NAMESPACE value: default $ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted $ kubectl apply -f deployment.yaml deployment.apps/emqx-deployment created $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-5884fd896f-b6hz9 1/1 Running 0 3s emqx-deployment-5884fd896f-hvd6v 1/1 Running 0 3s emqx-deployment-5884fd896f-jv4ck 1/1 Running 0 3s $ kubectl exec emqx-deployment-5884fd896f-b6hz9 -- emqx_ctl status Node 'emqx@172.20.32.168' is started emqx 4.1-rc.1 is running $ kubectl exec emqx-deployment-5884fd896f-b6hz9 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@172.20.32.168','emqx@172.20.58.237', 'emqx@172.20.85.197'], stopped_nodes => []} 中止一个 Pod：
$ kubectl delete pods emqx-deployment-5884fd896f-b6hz9 --force Warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely. pod "emqx-deployment-5884fd896f-b6hz9" force deleted $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-5884fd896f-84wxw 1/1 Running 0 34s emqx-deployment-5884fd896f-hvd6v 1/1 Running 0 5m11s emqx-deployment-5884fd896f-jv4ck 1/1 Running 0 5m11s $ kubectl exec emqx-deployment-5884fd896f-84wxw -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@172.20.32.169','emqx@172.20.58.237', 'emqx@172.20.85.197'], stopped_nodes => ['emqx@172.20.32.168']} 输出结果表明 EMQX Broker 会正确的显示已经停掉的 Pod，并将 Deployment 新建的 Pod 加入集群。至此，EMQX Broker 在 kubernetes 上成功建立集群。
7. 持久化 EMQX Broker 集群
上文中使用的 Deployment 来管理 Pod，但是 Pod 的网络是不停变动的，而且当 Pod 被销毁重建时，储存在 EMQX Broker 的数据和配置也就随之消失了，这在生产中是不能接受的，接下来尝试把 EMQX Broker 的集群持久化，即使 Pod 被销毁重建，EMQX Broker 的数据依然可以保存下来。
ConfigMap 是 configMap 是一种 API 对象，用来将非机密性的数据保存到健值对中。使用时可以用作环境变量、命令行参数或者存储卷中的配置文件。
ConfigMap 将您的环境配置信息和 容器镜像 解耦，便于应用配置的修改。
接下来使用 ConfigMap 记录 EMQX Broker 的配置，并将它们以环境变量的方式导入到 Deployment 中。定义 Configmap，并部署：
vim configmap.yaml apiVersion: v1 kind: ConfigMap metadata: name: emqx-config data: EMQX_CLUSTER__K8S__ADDRESS_TYPE: "hostname" EMQX_CLUSTER__K8S__APISERVER: "https://kubernetes.default.svc:443" EMQX_CLUSTER__K8S__SUFFIX: "svc.cluster.local" $ kubectl apply -f configmap.yaml configmap/emqx-config created 配置 Deployment 来使用 Configmap
$ vim deployment.yaml apiVersion: apps/v1 kind: Deployment metadata: name: emqx-deployment labels: app: emqx spec: replicas: 3 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 envFrom: - configMapRef: name: emqx-config 重新部署 Deployment，查看状态：
$ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted $ kubectl apply -f deployment.yaml deployment.apps/emqx-deployment created $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-5958799d95-7lfw6 1/1 Running 0 20s emqx-deployment-5958799d95-hfrvt 1/1 Running 0 20s emqx-deployment-5958799d95-q6d7p 1/1 Running 0 20s $ kubectl exec emqx-deployment-5958799d95-7lfw6 -- emqx_ctl status Node 'emqx-deployment-5958799d95-7lfw6@172.20.58.246' is started emqx 4.1-rc.1 is running $ kubectl exec emqx-deployment-5958799d95-7lfw6 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx-deployment-5958799d95-7lfw6@172.20.58.246'], stopped_nodes => []} 解决：
$ vim configmap.yaml apiVersion: v1 kind: ConfigMap metadata: name: emqx-config data: EMQX_NAME: "emqx" EMQX_CLUSTER__DISCOVERY: "k8s" EMQX_CLUSTER__K8S__APP_NAME: "emqx" EMQX_CLUSTER__K8S__SERVICE_NAME: "emqx-service" EMQX_CLUSTER__K8S__APISERVER: "https://kubernetes.default.svc:443" EMQX_CLUSTER__K8S__NAMESPACE: "default" $ kubectl delete configmap emqx-config configmap "emqx-config" deleted $ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-deployment-5958799d95-6vkth 1/1 Running 0 76m emqx-deployment-5958799d95-dtnwf 1/1 Running 0 76m emqx-deployment-5958799d95-j4vcf 1/1 Running 0 76m $ kubectl exec emqx-deployment-5958799d95-6vkth -- emqx_ctl status Node 'emqx@172.20.58.245' is started emqx 4.1-rc.1 is running $ kubectl exec emqx-deployment-5958799d95-6vkth -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@172.20.32.143','emqx@172.20.58.245', 'emqx@172.20.85.211'], stopped_nodes => []} EMQX Broker 的配置文件已经解耦到 Configmap 中了，如果有需要，可以自由的配置一个或多个 Configmap，并把它们作为环境变量或是文件引入到 Pod 内。
8. StatefulSet
StatefulSet 是为了解决有状态服务的问题（对应 Deployments 和 ReplicaSets 是为无状态服务而设计），其应用场景包括
- 稳定的持久化存储，即 Pod 重新调度后还是能访问到相同的持久化数据，基于 PVC 来实现
- 稳定的网络标志，即 Pod 重新调度后其 PodName 和 HostName 不变，基于 Headless Service（即没有Cluster IP的Service）来实现
- 稳定的网络标志，即 Pod 重新调度后其 PodName 和 HostName 不变，基于 Headless Service（即没有Cluster IP的Service）来实现
- 有序收缩，有序删除（即从N-1到0）
从上面的应用场景可以发现，StatefulSet由以下几个部分组成：
- 用于定义网络标志（DNS domain）的 Headless Service
- 用于创建 PersistentVolumes 的 volumeClaimTemplates
- 定义具体应用的 StatefulSet
StatefulSet 中每个 Pod 的 DNS 格式为 statefulSetName-{0..N-1}.serviceName.namespace.svc.cluster.local ，其中
- serviceName 为 Headless Service 的名字
0..N-1为 Pod 所在的序号，从 0 开始到 N-1- statefulSetName 为StatefulSet的名字
- namespace 为服务所在的 namespace，Headless Servic 和 StatefulSet 必须在相同的 namespace
.cluster.local为 Cluster Domain
接下来使用 StatefulSet 代替 Deployment 来管理 Pod。删除 Deployment：
$ kubectl delete deployment emqx-deployment deployment.apps "emqx-deployment" deleted 定义 StatefulSet：
vim statefulset.yaml apiVersion: apps/v1 kind: StatefulSet metadata: name: emqx-statefulset labels: app: emqx spec: serviceName: emqx-headless updateStrategy: type: RollingUpdate replicas: 3 selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 envFrom: - configMapRef: name: emqx-config 注意，StatefulSet 需要 Headless Service 来实现稳定的网络标志，因此需要再定义一个 Service
$ vim headless.yaml apiVersion: v1 kind: Service metadata: name: emqx-headless spec: type: ClusterIP clusterIP: None selector: app: emqx ports: - name: mqtt port: 1883 protocol: TCP targetPort: 1883 - name: mqttssl port: 8883 protocol: TCP targetPort: 8883 - name: mgmt port: 8081 protocol: TCP targetPort: 8081 - name: websocket port: 8083 protocol: TCP targetPort: 8083 - name: wss port: 8084 protocol: TCP targetPort: 8084 - name: dashboard port: 18083 protocol: TCP targetPort: 18083 因为 Headless Service 并不需要 IP，所以配置了 clusterIP: None 。部署相应的资源：
$ kubectl apply -f headless.yaml service/emqx-headless created $ kubectl get svc NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE emqx-headless ClusterIP None <none> 1883/TCP,8883/TCP,8081/TCP,8083/TCP,8084/TCP,18083/TCP 53s emqx-service ClusterIP 10.68.228.164 <none> 1883/TCP,8883/TCP,8081/TCP,8083/TCP,8084/TCP,18083/TCP 21h $ kubectl apply -f statefulset.yaml statefulset.apps/emqx-statefulset created $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 2m emqx-statefulset-1 1/1 Running 0 24s emqx-statefulset-2 1/1 Running 0 22s $ kubectl exec emqx-statefulset-0 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@172.20.32.174','emqx@172.20.58.248', 'emqx@172.20.85.214'], stopped_nodes => []} 更新 Configmap：StatefulSet 提供了稳定的网络标志，EMQX Broker 支持使用 hostname 和 dns 规则来代提 IP 实现集群，以 hostname 为例，需要修改 emqx.conf：
cluster.kubernetes.address_type = hostname cluster.kubernetes.suffix = svc.cluster.local kubernetes 集群中 Pod 的 DNS 规则可以由用户自定义，EMQX Broker 提供了 cluster.kubernetes.suffix 方便用户匹配自定的 DNS 规则，本文使用默认的 DNS 规则：statefulSetName-{0..N-1}.serviceName.namespace.svc.cluster.local ，DNS 规则中的 serviceName 为 StatefulSet 使用的 Headless Service，所以还需要将 cluster.kubernetes.service_name 修改为 Headless Service Name。
将配置项转为环境变量，需要在 Configmap 中配置：
EMQX_CLUSTER__K8S__ADDRESS_TYPE: "hostname" EMQX_CLUSTER__K8S__SUFFIX: "svc.cluster.local" EMQX_CLUSTER__K8S__SERVICE_NAME: emqx-headless Configmap 提供了热更新功能，执行 $ kubectl edit configmap emqx-config 来热更新 Configmap。
$ kubectl edit configmap emqx-config # 修改前 # Please edit the object below. Lines beginning with a '#' will be ignored, # and an empty file will abort the edit. If an error occurs while saving this file will be # reopened with the relevant failures. # apiVersion: v1 data: EMQX_CLUSTER__DISCOVERY: k8s EMQX_CLUSTER__K8S__APISERVER: https://kubernetes.default.svc:443 EMQX_CLUSTER__K8S__APP_NAME: emqx EMQX_CLUSTER__K8S__NAMESPACE: default EMQX_CLUSTER__K8S__SERVICE_NAME: emqx-service EMQX_NAME: emqx kind: ConfigMap metadata: annotations: kubectl.kubernetes.io/last-applied-configuration: | {"apiVersion":"v1","data":{"EMQX_CLUSTER__DISCOVERY":"k8s","EMQX_CLUSTER__K8S__APISERVER":"https://kubernetes.default.svc:443","EMQX_CLUSTER__K8S__APP_NAME":"emqx","EMQX_CLUSTER__K8S__NAMESPACE":"default","EMQX_CLUSTER__K8S__SERVICE_NAME":"emqx-service","EMQX_NAME":"emqx"},"kind":"ConfigMap","metadata":{"annotations":{},"name":"emqx-config","namespace":"default"}} creationTimestamp: "2024-07-12T05:26:16Z" name: emqx-config namespace: default resourceVersion: "1734159" uid: 13ccc86a-9ab3-450d-8a91-d73a7745d75b # 修改后 apiVersion: v1 data: EMQX_CLUSTER__DISCOVERY: k8s EMQX_CLUSTER__K8S__APISERVER: https://kubernetes.default.svc:443 EMQX_CLUSTER__K8S__APP_NAME: emqx EMQX_CLUSTER__K8S__NAMESPACE: default EMQX_CLUSTER__K8S__SERVICE_NAME: emqx-headless EMQX_NAME: emqx EMQX_CLUSTER__K8S__ADDRESS_TYPE: hostname EMQX_CLUSTER__K8S__SUFFIX: svc.cluster.local kind: ConfigMap metadata: annotations: kubectl.kubernetes.io/last-applied-configuration: | {"apiVersion":"v1","data":{"EMQX_CLUSTER__DISCOVERY":"k8s","EMQX_CLUSTER__K8S__APISERVER":"https://kubernetes.default.svc:443","EMQX_CLUSTER__K8S__APP_NAME":"emqx","EMQX_CLUSTER__K8S__NAMESPACE":"default","EMQX_CLUSTER__K8S__SERVICE_NAME":"emqx-headless","EMQX_NAME":"emqx","EMQX_CLUSTER__K8S__ADDRESS_TYPE":"hostname","EMQX_CLUSTER__K8S__SUFFIX":"svc.cluster.local"},"kind":"ConfigMap","metadata":{"annotations":{},"name":"emqx-config","namespace":"default"}} creationTimestamp: "2024-07-12T05:26:16Z" name: emqx-config namespace: default resourceVersion: "1734159" uid: 13ccc86a-9ab3-450d-8a91-d73a7745d75b 重新部署 StatefulSet：Configmap 更新之后 Pod 并不会重启，需要我们手动更新 StatefulSet
$ kubectl delete statefulset emqx-statefulset statefulset.apps "emqx-statefulset" deleted $ kubectl apply -f statefulset.yaml statefulset.apps/emqx-statefulset created $ kubectl exec emqx-statefulset-0 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@emqx-statefulset-0.emqx-headless.default.svc.cluster.local'], stopped_nodes => []} $ kubectl exec emqx-statefulset-0 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@emqx-statefulset-0.emqx-headless.default.svc.cluster.local'], stopped_nodes => []} $ kubectl logs pod/emqx-statefulset-0 -n default ... (emqx@emqx-statefulset-0.emqx-headless.default.svc.cluster.local)1> 2024-07-12 07:59:16.261 [error] Ekka(AutoCluster): Discover error: {badkey,<<"hostname">>} [{maps,get, [<<"hostname">>, #{<<"ip">> => <<"172.20.32.158">>,<<"nodeName">> => <<"master-01">>, <<"targetRef">> => #{<<"kind">> => <<"Pod">>,<<"name">> => <<"my-emqx-0">>, <<"namespace">> => <<"default">>, <<"uid">> => <<"de590aa1-d6da-42e0-9515-fe7f1fa68dfe">>}}], []}, {ekka_cluster_k8s,extract_host,2, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_cluster_k8s.erl"}, {line,114}]}, {ekka_cluster_k8s,'-extract_addresses/2-lc$^1/1-1-',2, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_cluster_k8s.erl"}, {line,110}]}, {ekka_cluster_k8s,'-extract_addresses/2-lc$^1/1-1-',2, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_cluster_k8s.erl"}, {line,111}]}, {ekka_cluster_k8s,'-extract_addresses/2-lc$^0/1-0-',2, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_cluster_k8s.erl"}, {line,111}]}, {ekka_cluster_k8s,extract_addresses,2, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_cluster_k8s.erl"}, {line,112}]}, {ekka_cluster_k8s,discover,1, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_cluster_k8s.erl"}, {line,48}]}, {ekka_autocluster,discover_and_join,2, [{file,"/emqx_rel/_build/emqx/lib/ekka/src/ekka_autocluster.erl"}, {line,125}]}] 原因：我当时是已经存在一套 emqx 集群在 k8s 上且在正常使用，将这套 my-emqx 的删除即可。
$ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 115s emqx-statefulset-1 1/1 Running 0 112s emqx-statefulset-2 1/1 Running 0 110s $ kubectl exec emqx-statefulset-2 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@emqx-statefulset-0.emqx-headless.default.svc.cluster.local', 'emqx@emqx-statefulset-1.emqx-headless.default.svc.cluster.local', 'emqx@emqx-statefulset-2.emqx-headless.default.svc.cluster.local'], stopped_nodes => []} 注意：当改为版本 5.7.1 版本的时候，修改 statefulset.yaml 相关内容为 image: emqx/emqx:5.7.1；configmap.yaml 文件里的配置参数名称有变化，需要改为
apiVersion: v1 kind: ConfigMap metadata: name: emqx-config data: EMQX_NAME: "emqx" EMQX_CLUSTER__DISCOVERY_STRATEGY: "k8s" EMQX_CLUSTER__K8S__APISERVER: "https://kubernetes.default.svc:443" EMQX_CLUSTER__K8S__SERVICE_NAME: "emqx-headless" EMQX_CLUSTER__K8S__NAMESPACE: "default" EMQX_CLUSTER__K8S__ADDRESS_TYPE: "hosname" EMQX_CLUSTER__K8S__SUFFIX: "svc.cluster.local" 可以看到新的 EMQX Broker 集群已经成功的建立起来了。中止一个 Pod：StatefulSet 中的 Pod 重新调度后其 PodName 和 HostName 不变，下面来尝试一下：
$ kubectl get pods kuNAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 6m20s emqx-statefulset-1 1/1 Running 0 6m17s emqx-statefulset-2 1/1 Running 0 6m15s $ kubectl delete pod emqx-statefulset-0 pod "emqx-statefulset-0" deleted $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 27s emqx-statefulset-1 1/1 Running 0 9m45s emqx-statefulset-2 1/1 Running 0 9m43s $ kubectl exec emqx-statefulset-2 -- emqx_ctl cluster status Cluster status: #{running_nodes => ['emqx@emqx-statefulset-0.emqx-headless.default.svc.cluster.local', 'emqx@emqx-statefulset-1.emqx-headless.default.svc.cluster.local', 'emqx@emqx-statefulset-2.emqx-headless.default.svc.cluster.local'], stopped_nodes => []} 跟预期的一样，StatefulSet 重新调度了一个具有相同网络标志的 Pod，Pod 中的 EMQX Broker 也成功的加入了集群。
9. StorageClasses、PersistentVolume 和 PersistentVolumeClaim
PersistentVolume（PV）是由管理员设置的存储，它是群集的一部分。就像节点是集群中的资源一样，PV 也是集群中的资源。 PV 是 Volume 之类的卷插件，但具有独立于使用 PV 的 Pod 的生命周期。此 API 对象包含存储实现的细节，即 NFS、iSCSI 或特定于云供应商的存储系统。
PersistentVolumeClaim（PVC）是用户存储的请求。它与 Pod 相似。Pod 消耗节点资源，PVC 消耗 PV 资源。Pod 可以请求特定级别的资源（CPU 和内存）。声明可以请求特定的大小和访问模式（例如，可以以读/写一次或 只读多次模式挂载）。
StorageClass 为管理员提供了描述存储 “class（类）” 的方法。 不同的 class 可能会映射到不同的服务质量等级或备份策略，或由群集管理员确定的任意策略。 Kubernetes 本身不清楚各种 class 代表的什么。这个概念在其他存储系统中有时被称为“配置文件”。
在部署 EMQX Broker 的时候，可以预先创建好 PV 或 StorageClass，然后利用 PVC 将 EMQX Broker 的 /opt/emqx/data/mnesia 目录挂载出来，当Pods被重新调度之后，EMQX 会从 /opt/emqx/data/mnesia 目录中获取数据并恢复，从而实现 EMQX Broker 的持久化。
定义 StatefulSet：
$ vim statefulset.yaml apiVersion: apps/v1 kind: StatefulSet metadata: name: emqx-statefulset labels: app: emqx spec: replicas: 3 serviceName: emqx-headless updateStrategy: type: RollingUpdate selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: volumes: - name: emqx-data persistentVolumeClaim: claimName: emqx-pvc serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 envFrom: - configMapRef: name: emqx-config volumeMounts: - name: emqx-data mountPath: "/opt/emqx/data/mnesia" volumeClaimTemplates: - metadata: name: emqx-pvc annotations: volume.alpha.kubernetes.io/storage-class: manual spec: accessModes: [ "ReadWriteOnce" ] resources: requests: storage: 1Gi 该文件首先通过 volumeClaimTemplates 指定了使用 StorageClass 的 name 为 manual 的存储类创建名称为 emqx-pvc 的 PVC 资源，PVC 资源的读写模式为 ReadWriteOnce，需要 1Gi 的空间，然后将此 PVC 定义为 name 为 emqx-data 的 volumes，并将此 volumes 挂载在 Pod 中的 /opt/emqx/data/mnesia 目录下。
$ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc-emqx-statefulset-0 Pending $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 0/1 Pending 0 75s $ kubectl describe pvc emqx-pvc-emqx-statefulset-0 Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal FailedBinding 4s (x2 over 15s) persistentvolume-controller no persistent volumes available for this claim and no storage class is set $ kubectl describe pod emqx-statefulset-0 Events: Type Reason Age From Message ---- ------ ---- ---- ------- Warning FailedScheduling 109s default-scheduler 0/3 nodes are available: persistentvolumeclaim "local-pv-sdc" not found. preemption: 0/3 nodes are available: 3 No preemption victims found for incoming pod.. 解决：手动创建 emqx-pvc。参考：在K8S中使用Local持久卷
$ kubectl delete statefulset emqx-statefulset statefulset.apps "emqx-statefulset" deleted $ kubectl delete pvc emqx-pvc-emqx-statefulset-0 persistentvolumeclaim "emqx-pvc-emqx-statefulset-0" deleted $ vim emqx-pvc.yml kind: PersistentVolumeClaim apiVersion: v1 metadata: name: emqx-pvc spec: accessModes: - ReadWriteOnce storageClassName: local-redis-storage resources: requests: storage: 1Gi $ kubectl create -f emqx-pvc.yml persistentvolumeclaim/emqx-pvc created $ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc Pending local-redis-storage 5s $ vim statefulset.yaml apiVersion: apps/v1 kind: StatefulSet metadata: name: emqx-statefulset labels: app: emqx spec: replicas: 3 serviceName: emqx-headless updateStrategy: type: RollingUpdate selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: volumes: - name: emqx-data persistentVolumeClaim: claimName: emqx-pvc serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 envFrom: - configMapRef: name: emqx-config volumeMounts: - name: emqx-data mountPath: "/opt/emqx/data/mnesia" $ kubectl apply -f statefulset.yaml statefulset.apps/emqx-statefulset created $ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc Bound pvc-3fd906c3-5372-4e92-b971-062bc235b268 1Gi RWO local-redis-storage 22s $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 11s emqx-statefulset-1 1/1 Running 0 6s emqx-statefulset-2 1/1 Running 0 5s 总感觉原文章里配置的有问题，如果想有三个 pvc 的话也得手动创建，无法像原文章里那样，而且他们的前置条件是 emqx-pvc 已经存在。参考：helm或者k8s部署pod时遇到pod一直处于pending状态
$ kubectl get pvc emqx-pvc-emqx-statefulset-0 -o yaml > emqx-pvc-emqx-statefulset-0.yaml $ vim emqx-pvc-emqx-statefulset-0.yaml apiVersion: v1 kind: PersistentVolumeClaim metadata: annotations: volume.alpha.kubernetes.io/storage-class: local-redis-storage creationTimestamp: "2024-07-18T05:09:03Z" finalizers: - kubernetes.io/pvc-protection labels: app: emqx name: emqx-pvc-emqx-statefulset-0 namespace: default resourceVersion: "3204901" uid: 74f369fb-08d0-41a7-8230-83fddb0efe50 spec: accessModes: - ReadWriteOnce resources: requests: storage: 1Gi volumeMode: Filesystem status: phase: Pending # 添加如下内容： spec: storageClassName: local-redis-storage $ kubectl create -f emqx-pvc-emqx-statefulset-0.yaml persistentvolumeclaim/emqx-pvc-emqx-statefulset-0 created $ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc-emqx-statefulset-0 Pending local-redis-storage 5s $ kubectl get pvc emqx-pvc-emqx-statefulset-0 -o yaml > emqx-pvc-emqx-statefulset-1.yaml # 同样添加 storageClassName: local-redis-storage 并修改 name 为 emqx-pvc-emqx-statefulset-1 $ kubectl create -f emqx-pvc-emqx-statefulset-1.yaml $ kubectl get pvc emqx-pvc-emqx-statefulset-0 -o yaml > emqx-pvc-emqx-statefulset-2.yaml # 同样添加 storageClassName: local-redis-storage 并修改 name 为 emqx-pvc-emqx-statefulset-2 $ kubectl create -f emqx-pvc-emqx-statefulset-2.yaml $ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc Bound pvc-c12d4f80-07c8-4c72-8a97-765f74f34d53 1Gi RWO local-redis-storage 50m emqx-pvc-emqx-statefulset-0 Pending local-redis-storage 5m22s emqx-pvc-emqx-statefulset-1 Pending local-redis-storage 3s emqx-pvc-emqx-statefulset-2 Pending local-redis-storage 70s $ vim statefulset.yaml apiVersion: apps/v1 kind: StatefulSet metadata: name: emqx-statefulset labels: app: emqx spec: replicas: 3 serviceName: emqx-headless updateStrategy: type: RollingUpdate selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: volumes: - name: emqx-data persistentVolumeClaim: claimName: emqx-pvc serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 envFrom: - configMapRef: name: emqx-config volumeMounts: - name: emqx-data mountPath: "/opt/emqx/data/mnesia" volumeClaimTemplates: - metadata: name: emqx-pvc annotations: volume.alpha.kubernetes.io/storage-class: local-redis-storage spec: accessModes: [ "ReadWriteOnce" ] resources: requests: storage: 1Gi $ kubectl apply -f statefulset.yaml statefulset.apps/emqx-statefulset created $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 2m3s emqx-statefulset-1 1/1 Running 0 117s emqx-statefulset-2 1/1 Running 0 112s $ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc Bound pvc-c12d4f80-07c8-4c72-8a97-765f74f34d53 1Gi RWO local-redis-storage 54m emqx-pvc-emqx-statefulset-0 Bound pvc-a718e44c-0b2d-433c-be49-d264eb44b878 1Gi RWO local-redis-storage 8m39s emqx-pvc-emqx-statefulset-1 Bound pvc-1bb39cb2-4a23-49ab-b874-756cd1a2502f 1Gi RWO local-redis-storage 3m20s emqx-pvc-emqx-statefulset-2 Bound pvc-7df2cdf4-fc81-47f8-b02b-ffa8e8bb0ae6 1Gi RWO local-redis-storage 4m27s 综上所述，也可以不用手动创建 emqx-pvc-emqx-statefulset-0、emqx-pvc-emqx-statefulset-1、emqx-pvc-emqx-statefulset-2，其实就是在 statefulset.yaml 中加一行配置的事，最终的文件内容为：
$ vim statefulset.yaml apiVersion: apps/v1 kind: StatefulSet metadata: name: emqx-statefulset labels: app: emqx spec: replicas: 3 serviceName: emqx-headless updateStrategy: type: RollingUpdate selector: matchLabels: app: emqx template: metadata: labels: app: emqx spec: volumes: - name: emqx-data persistentVolumeClaim: claimName: emqx-pvc serviceAccountName: emqx containers: - name: emqx image: emqx/emqx:v4.1-rc.1 ports: - name: mqtt containerPort: 1883 - name: mqttssl containerPort: 8883 - name: mgmt containerPort: 8081 - name: ws containerPort: 8083 - name: wss containerPort: 8084 - name: dashboard containerPort: 18083 envFrom: - configMapRef: name: emqx-config volumeMounts: - name: emqx-data mountPath: "/opt/emqx/data/mnesia" volumeClaimTemplates: - metadata: name: emqx-pvc annotations: volume.alpha.kubernetes.io/storage-class: local-redis-storage spec: storageClassName: local-redis-storage accessModes: [ "ReadWriteOnce" ] resources: requests: storage: 1Gi $ kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc Bound pvc-ce861efc-751c-4fd5-b96f-2605fb41e7d2 1Gi RWO local-redis-storage 15m $ kubectl apply -f statefulset.yaml statefulset.apps/emqx-statefulset created kubectl get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE emqx-pvc Bound pvc-ce861efc-751c-4fd5-b96f-2605fb41e7d2 1Gi RWO local-redis-storage 15m emqx-pvc-emqx-statefulset-0 Bound pvc-21a89457-5c05-4df3-b447-c7fad1424df8 1Gi RWO local-redis-storage 3m39s emqx-pvc-emqx-statefulset-1 Bound pvc-234da635-d316-4256-b559-98dab56cc5a4 1Gi RWO local-redis-storage 3m33s emqx-pvc-emqx-statefulset-2 Bound pvc-ec3bcfee-27ac-4776-8662-aa504676ca6a 1Gi RWO local-redis-storage 3m28s $ kubectl get pods NAME READY STATUS RESTARTS AGE emqx-statefulset-0 1/1 Running 0 5m43s emqx-statefulset-1 1/1 Running 0 5m37s emqx-statefulset-2 1/1 Running 0 5m32s 输出结果表明该 PVC 的状态为 Bound，PVC 存储已经成功的建立了，当 Pod 被重新调度时，EMQX Broker 会读取挂载到 PVC 中的数据，从而实现持久化。
