云原生管理端设备

本文档主要讲述如何在已有的OpenYurt集群上安装Yurt-Device-Controller 和 Yurt-EdgeX-Manager组件，并通过部署虚拟端设备来展示如何通过云原生的方式管理边缘端设备。

对于有兴趣的读者，可以去相关的github仓库参考本文使用组件的具体实现：Yurt-Device-Controller 和 Yurt-EdgeX-Manager

如果你还没有OpenYurt集群，你可以使用 yurtctl工具 来初始化一个OpenYurt集群或将一个Kubernetes集群转换为OpenYurt集群。

环境要求

  • OpenYurt v0.5.0+

  • 安装了 Yurt-app-manager 组件

  • 与master不在同一局域网下的节点都需要部署coreDNS pod

  • 将访问coreDNS service的流量改为节点池内闭环，参考教程

1. 安装yurt-edgex-manager并创建一个EdgeX实例

安装部署yurt-edgex-manager

  1. # 如果期望部署edgex的节点是arm64架构，则使用以下的yaml文件
  2. # kubectl apply -f https://raw.githubusercontent.com/openyurtio/yurt-edgex-manager/main/Documentation/yurt-edgex-manager-arm64.yaml
  3. $ kubectl apply -f https://raw.githubusercontent.com/openyurtio/yurt-edgex-manager/main/Documentation/yurt-edgex-manager.yaml
  5. # 检查状态
  6. $ kubectl get pods -n edgex-system |grep edgex
  7. edgex-controller-manager-6c99fd9f9f-b9nnk   2/2     Running   0          6d22h

创建一个hangzhou边缘节点池，并将边缘节点加入到hangzhou节点池

  1. $ export WORKER_NODEPOOL="hangzhou"
  2. $ export EDGE_NODE="node1"
  4. # 创建hangzhou节点池
  5. $ cat <<EOF | kubectl apply -f -
  6. apiVersion: apps.openyurt.io/v1alpha1
  7. kind: NodePool
  8. metadata:
  9.   name: $WORKER_NODEPOOL
  10. spec:
  11.   type: Edge
  12. EOF
  14. # 将边缘节点加入hangzhou节点池
  15. $ kubectl label node $EDGE_NODE apps.openyurt.io/desired-nodepool=hangzhou
  17. # 检查节点池状态
  18. $ kubectl get nodepool
  19. NAME      TYPE   READYNODES   NOTREADYNODES   AGE
  20. hangzhou   Edge   0            1               6d22h

在hangzhou节点池中创建edgex foundry实例，并在edgex中部署虚拟设备edgex-device-virtual

  1. $ export WORKER_NODEPOOL="hangzhou"
  2. # 如果部署节点是arm64，则改为"edgexfoundry/docker-device-virtual-go-arm64:1.3.0"
  3. $ export VIRTUAL_DEVICE_IMAGE="edgexfoundry/docker-device-virtual-go:1.3.0"
  5. $ cat <<EOF | kubectl apply -f -
  6. apiVersion: device.openyurt.io/v1alpha1
  7. kind: EdgeX
  8. metadata:
  9.   name: edgex-sample-$WORKER_NODEPOOL
  10. spec:
  11.   version: hanoi
  12.   poolname: $WORKER_NODEPOOL
  13.   additinalservices:
  14.   - metadata:
  15.       name: edgex-device-virtual
  16.     spec:
  17.       type: NodePort
  18.       selector:
  19.         app: edgex-device-virtual
  20.       ports:
  21.       - name: http
  22.         port: 49990
  23.         protocol: TCP
  24.         targetPort: 49990
  25.         nodePort: 30090
  26.   additinaldeployments:
  27.   - metadata:
  28.       name: edgex-device-virtual
  29.     spec:
  30.       selector:
  31.         matchLabels:
  32.           app: edgex-device-virtual
  33.       template:
  34.         metadata:
  35.           labels:
  36.             app: edgex-device-virtual
  37.         spec:
  38.           hostname: edgex-device-virtual
  39.           containers:
  40.           - name: edgex-device-virtual
  41.             image: $VIRTUAL_DEVICE_IMAGE
  42.             imagePullPolicy: IfNotPresent
  43.             ports:
  44.             - name: http
  45.               protocol: TCP
  46.               containerPort: 49990
  47.             envFrom:
  48.             - configMapRef:
  49.                 name: common-variables
  50.             env:
  51.               - name: Service_Host
  52.                 value: "edgex-device-virtual"
  53. EOF

检查edgex-foundry的部署状态

  1. $ kubectl get edgex
  2. NAME                    READY   SERVICE   READYSERVICE   DEPLOYMENT   READYDEPLOYMENT
  3. edgex-sample-hangzhou   true    9         9              9            9

2. 安装并部署yurt-device-controller

安装yurt-device-controller 相关的CRD

  1. $ kubectl apply -f https://raw.githubusercontent.com/openyurtio/yurt-device-controller/main/config/setup/crd.yaml

使用UnitedDeployment在hanghzou节点池中部署一个yurt-device-controller实例

  1. $ export WORKER_NODEPOOL="hangzhou"
  2. $ cat <<EOF | kubectl apply -f -
  3. apiVersion: apps.openyurt.io/v1alpha1
  4. kind: UnitedDeployment
  5. metadata:
  6.   labels:
  7.     controller-tools.k8s.io: "1.0"
  8.   name: yurt-device-controller
  9. spec:
  10.   selector:
  11.     matchLabels:
  12.       app: yurt-device-controller
  13.   workloadTemplate:
  14.     deploymentTemplate:
  15.       metadata:
  16.         labels:
  17.           app: yurt-device-controller
  18.       spec:
  19.         template:
  20.           metadata:
  21.             labels:
  22.               app: yurt-device-controller
  23.               control-plane: controller-manager
  24.           spec:
  25.             containers:
  26.             - args:
  27.               - --health-probe-bind-address=:8081
  28.               - --metrics-bind-address=127.0.0.1:8080
  29.               - --leader-elect=false
  30.               command:
  31.               - /yurt-device-controller
  32.               image: openyurt/yurt-device-controller:latest
  33.               imagePullPolicy: IfNotPresent
  34.               livenessProbe:
  35.                 httpGet:
  36.                   path: /healthz
  37.                   port: 8081
  38.                 initialDelaySeconds: 15
  39.                 periodSeconds: 20
  40.               name: manager
  41.               readinessProbe:
  42.                 httpGet:
  43.                   path: /readyz
  44.                   port: 8081
  45.                 initialDelaySeconds: 5
  46.                 periodSeconds: 10
  47.               resources:
  48.                 limits:
  49.                   cpu: 100m
  50.                   memory: 30Mi
  51.                 requests:
  52.                   cpu: 100m
  53.                   memory: 20Mi
  54.               securityContext:
  55.                 allowPrivilegeEscalation: false
  56.             securityContext:
  57.               runAsUser: 65532
  58.             terminationGracePeriodSeconds: 10
  59.   topology:
  60.     pools:
  61.     - name: $WORKER_NODEPOOL
  62.       nodeSelectorTerm:
  63.         matchExpressions:
  64.         - key: apps.openyurt.io/nodepool
  65.           operator: In
  66.           values:
  67.           - $WORKER_NODEPOOL
  68.       replicas: 1
  69.       tolerations:
  70.       - effect: NoSchedule
  71.         key: apps.openyurt.io/example
  72.         operator: Exists
  73.   revisionHistoryLimit: 5
  74. ---
  75. kind: ClusterRoleBinding
  76. apiVersion: rbac.authorization.k8s.io/v1beta1
  77. metadata:
  78.   name: default-cluster-admin
  79. subjects:
  80. - kind: ServiceAccount
  81.   name: default
  82.   namespace: default
  83. roleRef:
  84.   kind: ClusterRole
  85.   name: cluster-admin
  86.   apiGroup: ""
  87. ---
  88. EOF

检查yurt-device-controller是否部署成功

  1. $ kubectl get pod |grep yurt-device-controller
  2. yurt-device-controller-beijing-sf7xz-79c9cbf4b7-mbfds             1/1     Running   0          6d22h

3. 查看同步上来设备相关信息

可以通过以下命令查看同步上来设备相关信息

  1. $ kubectl get device
  2. $ kubectl get deviceservice
  3. $ kubectl get deviceprofile

4. 卸载相关组件并清理环境

  1. $ export WORKER_NODEPOOL="hangzhou"
  2. $ export EDGE_NODE="node1"
  4. # 1.1 删除所有device, deviceservice, deviceprofile资源
  5. $ kubectl delete device --all
  6. $ kubectl delete deviceprofile --all
  7. $ kubectl delete deviceservice --all
  9. # 1.2 删除部署的yurt-device-controller
  10. $ kubectl delete uniteddeployment yurt-device-controller
  11. $ kubectl delete clusterrolebinding default-cluster-admin
  13. # 1.3 删除device, deviceservice, deviceprofile资源相关的crd
  14. $ kubectl delete -f https://raw.githubusercontent.com/openyurtio/yurt-device-controller/main/config/setup/crd.yaml
  16. # 2.1 删除所有edgex实例
  17. $ kubectl delete edgex --all
  19. # 2.2 卸载yurt-edgex-manager
  20. # 如果使用的arm64版本的，使用以下命令
  21. # kubectl delete -f https://raw.githubusercontent.com/openyurtio/yurt-edgex-manager/main/Documentation/yurt-edgex-manager-arm64.yaml
  22. $ kubectl delete -f https://raw.githubusercontent.com/openyurtio/yurt-edgex-manager/main/Documentation/yurt-edgex-manager.yaml
  24. # （以下步骤可选）
  25. # 3.1 将边缘节点移除hangzhou节点池
  26. $ kubectl label node $EDGE_NODE apps.openyurt.io/desired-nodepool-
  27. # 3.2 删除hangzhou节点池
  28. $ kubectl delete nodepool $WORKER_NODEPOOL