Kubernetes master 不附加 FlexVolume

Question

我正在尝试为 Kubernetes 附加 dummy-attachable FlexVolume sample，根据我在节点和主节点上的日志，它似乎正常初始化：

Loaded volume plugin "flexvolume-k8s/dummy-attachable

但是当我尝试将卷附加到 pod 时，主节点永远不会调用附加方法。来自节点的日志读取：

flexVolume driver k8s/dummy-attachable: using default GetVolumeName for volume dummy-attachable
operationExecutor.VerifyControllerAttachedVolume started for volume "dummy-attachable"
Operation for "\"flexvolume-k8s/dummy-attachable/dummy-attachable\"" failed. No retries permitted until 2019-04-22 13:42:51.21390334 +0000 UTC m=+4814.674525788 (durationBeforeRetry 500ms). Error: "Volume has not been added to the list of VolumesInUse in the node's volume status for volume \"dummy-attachable\" (UniqueName: \"flexvolume-k8s/dummy-attachable/dummy-attachable\") pod \"nginx-dummy-attachable\"

这是我尝试装载卷的方式：

apiVersion: v1
kind: Pod
metadata:
  name: nginx-dummy-attachable
  namespace: default
spec:
  containers:
    - name: nginx-dummy-attachable
      image: nginx
      volumeMounts:
        - name: dummy-attachable
          mountPath: /data
      ports:
        - containerPort: 80
  volumes:
    - name: dummy-attachable
      flexVolume:
        driver: "k8s/dummy-attachable"

这是 kubectl describe pod nginx-dummy-attachable 的输出：

Name:               nginx-dummy-attachable
Namespace:          default
Priority:           0
PriorityClassName:  <none>
Node:               [node id]
Start Time:         Wed, 24 Apr 2019 08:03:21 -0400
Labels:             <none>
Annotations:        kubernetes.io/limit-ranger: LimitRanger plugin set: cpu request for container nginx-dummy-attachable
Status:             Pending
IP:                 
Containers:
  nginx-dummy-attachable:
    Container ID:   
    Image:          nginx
    Image ID:       
    Port:           80/TCP
    Host Port:      0/TCP
    State:          Waiting
      Reason:       ContainerCreating
    Ready:          False
    Restart Count:  0
    Requests:
      cpu:        100m
    Environment:  <none>
    Mounts:
      /data from dummy-attachable (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-hcnhj (ro)
Conditions:
  Type              Status
  Initialized       True 
  Ready             False 
  ContainersReady   False 
  PodScheduled      True 
Volumes:
  dummy-attachable:
    Type:       FlexVolume (a generic volume resource that is provisioned/attached using an exec based plugin)
    Driver:     k8s/dummy-attachable
    FSType:     
    SecretRef:  nil
    ReadOnly:   false
    Options:    map[]
  default-token-hcnhj:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-hcnhj
    Optional:    false
QoS Class:       Burstable
Node-Selectors:  <none>
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type     Reason       Age                From                                    Message
  ----     ------       ----               ----                                    -------
  Warning  FailedMount  41s (x6 over 11m)  kubelet, [node id]  Unable to mount volumes for pod "nginx-dummy-attachable_default([id])": timeout expired waiting for volumes to attach or mount for pod "default"/"nginx-dummy-attachable". list of unmounted volumes=[dummy-attachable]. list of unattached volumes=[dummy-attachable default-token-hcnhj]

我向 FlexVolume 添加了调试日志记录，因此我能够验证从未在主节点上调用附加方法。我不确定我在这里遗漏了什么。

我不知道这是否重要，但是正在使用 KOPS 启动集群。我试过 k8s 1.11 和 1.14 都没有成功。

Answer 1

所以这很有趣。

尽管 kubelet 在 master 上初始化了 FlexVolume 插件，但 kube-controller-manager 是容器化在 KOP 中的，它是实际负责将卷附加到 pod 的应用程序。 KOPs 不会将默认插件目录 /usr/libexec/kubernetes/kubelet-plugins/volume/exec 挂载到 kube-controller-manager pod 中，因此它对 master 上的 FlexVolume 插件一无所知。

在 KOPs 解决这个问题之前，除了使用不同的 Kubernetes 部署工具之外，似乎没有一种简单的方法可以做到这一点。

Kubernetes master 不附加 FlexVolume

Kubernetes master doesn't attach FlexVolume

kubernetes

kops