K3S 集群在 Rancher 仪表板中处于待处理状态

Question

我已经用 K3S 安装了一个 3 节点集群。 kubectl 正确检测到节点，我能够部署图像。

$ k3s kubectl get nodes
NAME                   STATUS   ROLES                       AGE     VERSION
master                 Ready    control-plane,etcd,master   4h31m   v1.22.2+k3s1
worker-01              Ready    <none>                      3h59m   v1.22.2+k3s1
worker-02              Ready    <none>                      4h3m    v1.22.2+k3s1

我还安装了 Rancher 最新版本（2.6.0) 通过 docker-compose:

version: '2'
services:
  rancher:
    image: rancher/rancher:latest
    restart: always
    ports:
    - "8080:80/tcp"
    - "4443:443/tcp"
    volumes:
    - "rancher-data:/var/lib/rancher"
    privileged: true
volumes:
  rancher-data:

仪表板可从每个节点访问，我已经导入了一个现有集群，运行以下命令：

curl --insecure -sfL https://192.168.1.100:4443/v3/import/66txfzmv4fnw6bqj99lpmdt6jlx4rpwblzhx96wvljc8gczphcn2c2_c-m-nz826pgl.yaml | kubectl apply -f -

集群显示为 Active 但有 0 个节点和消息：

[Pending] waiting for full cluster configuration

完整的 yaml 状态在这里：

apiVersion: provisioning.cattle.io/v1
kind: Cluster
metadata:
  annotations:
    field.cattle.io/creatorId: user-5bk6w
  creationTimestamp: "2021-10-05T10:06:35Z"
  finalizers:
  - wrangler.cattle.io/provisioning-cluster-remove
  generation: 1
  managedFields:
  - apiVersion: provisioning.cattle.io/v1
    fieldsType: FieldsV1
    fieldsV1:
      f:metadata:
        f:finalizers:
          .: {}
          v:"wrangler.cattle.io/provisioning-cluster-remove": {}
      f:spec: {}
      f:status:
        .: {}
        f:clientSecretName: {}
        f:clusterName: {}
        f:conditions: {}
        f:observedGeneration: {}
        f:ready: {}
    manager: rancher
    operation: Update
    time: "2021-10-05T10:08:30Z"
  name: ofb
  namespace: fleet-default
  resourceVersion: "73357"
  uid: 1d03f05e-77b7-4361-947d-2ef5b50928f5
spec: {}
status:
  clientSecretName: ofb-kubeconfig
  clusterName: c-m-nz826pgl
  conditions:
  - lastUpdateTime: "2021-10-05T10:08:30Z"
    status: "False"
    type: Reconciling
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "False"
    type: Stalled
  - lastUpdateTime: "2021-10-05T14:08:52Z"
    status: "True"
    type: Created
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: RKECluster
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: BackingNamespaceCreated
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: DefaultProjectCreated
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: SystemProjectCreated
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: InitialRolesPopulated
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: CreatorMadeOwner
  - lastUpdateTime: "2021-10-05T10:08:15Z"
    status: "True"
    type: Pending
  - lastUpdateTime: "2021-10-05T10:08:15Z"
    message: waiting for full cluster configuration
    reason: Pending
    status: "True"
    type: Provisioned
  - lastUpdateTime: "2021-10-05T14:08:52Z"
    message: Waiting for API to be available
    status: "True"
    type: Waiting
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: NoDiskPressure
  - lastUpdateTime: "2021-10-05T10:06:35Z"
    status: "True"
    type: NoMemoryPressure
  - lastUpdateTime: "2021-10-05T10:06:39Z"
    status: "False"
    type: Connected
  - lastUpdateTime: "2021-10-05T14:04:52Z"
    status: "True"
    type: Ready
  observedGeneration: 1
  ready: true

集群代理未显示任何特殊问题：

$ kubectl -n cattle-system logs -l app=cattle-cluster-agent
time="2021-10-05T13:54:30Z" level=info msg="Connecting to wss://192.168.1.100:4443/v3/connect with token starting with 66txfzmv4fnw6bqj99lpmdt6jlx"
time="2021-10-05T13:54:30Z" level=info msg="Connecting to proxy" url="wss://192.168.1.100:4443/v3/connect"

我需要做些什么才能使集群完全运行吗？我试图将 Rancher 版本降级到 2.5.0 但我遇到了同样的问题。

Answer 1

这个问题通常是因为 cattle-cluster-agent 无法连接到配置的服务器-url。另外，让我告诉您，从 Rancher 2.5 和更新版本开始，cattle-node-agents 仅存在于使用 RKE 在 Rancher 中创建的集群中。您可以访问以下 URL，其中包含与您的问题相关的官方 Rancher 文档，并按照那里的说明解决问题：

Rancher Registered clusters

Answer 2

我认为这与 Kubernetes v1.22 不兼容。

在导入新的 v1.22.2 集群时遇到与 Rancher v2.6.0 相同的问题（运行在 IBM Cloud VPC 基础设施上），我跟踪了 Docker 容器的日志运行牧场主并观察到：

2021/10/20 14:40:31 [INFO] Starting cluster controllers for c-m-cs78tnxc
E1020 14:40:31.346373      33 reflector.go:139] pkg/mod/github.com/rancher/client-go@v0.21.0-rancher.1/tools/cache/reflector.go:168: Failed to watch *v1beta1.Ingress: failed to list *v1beta1.Ingress: the server could not find the requested resource (get ingresses.meta.k8s.io)

Kubernetes v1.22 updates NGINX-Ingress to v1.x, which appears to be the cause, and there is an open issue on the Rancher GitHub 更新它以与 Kubernetes v1.22 兼容。

同时，在同一基础设施上使用 Kubernetes v1.21.5 重新创建新集群后，我能够成功将其导入 Rancher。

K3S 集群在 Rancher 仪表板中处于待处理状态

K3S cluster is pending in Rancher dashboard

kubernetes

rancher

kubectl

k3s