MountVolume.MountDevice failed for volume / Timeout waiting for mount

[wccropper]

I am trying to use persistent manual bucket and share it across pods:

kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
  name: bucket-csi-s3
provisioner: ru.yandex.s3.csi
parameters:
  mounter: geesefs
  options: "--memory-limit 1000 --dir-mode 0777 --file-mode 0666"
  bucket: csi-s3-1
  endpoint: https://<company.internal.s3.endpoint>    # actual SSL cert, not self-signed
  csi.storage.k8s.io/provisioner-secret-name: csi-s3-secret
  csi.storage.k8s.io/provisioner-secret-namespace: csi-s3
  csi.storage.k8s.io/controller-publish-secret-name: csi-s3-secret
  csi.storage.k8s.io/controller-publish-secret-namespace: csi-s3
  csi.storage.k8s.io/node-stage-secret-name: csi-s3-secret
  csi.storage.k8s.io/node-stage-secret-namespace: csi-s3
  csi.storage.k8s.io/node-publish-secret-name: csi-s3-secret
  csi.storage.k8s.io/node-publish-secret-namespace: csi-s3
reclaimPolicy: Retain

apiVersion: v1
kind: PersistentVolume
metadata:
  name: manualbucket-with-path
spec:
  storageClassName: bucket-csi-s3
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteMany
  claimRef:
    namespace: csi-s3
    name: bucket-csi-s3-manual-pvc
  csi:
    driver: ru.yandex.s3.csi
    controllerPublishSecretRef:
      name: csi-s3-secret
      namespace: csi-s3
    nodePublishSecretRef:
      name: csi-s3-secret
      namespace: csi-s3
    nodeStageSecretRef:
      name: csi-s3-secret
      namespace: csi-s3
    volumeAttributes:
      capacity: 10Gi
      mounter: geesefs
      options: --memory-limit 1000 --dir-mode 0777 --file-mode 0666
    volumeHandle: manualbucket/path

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: bucket-csi-s3-manual-pvc
spec:
  storageClassName: ""
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: 10Gi

apiVersion: v1
kind: Pod
metadata:
  name: csi-s3-test-nginx
  namespace: csi-s3
spec:
  containers:
   - name: csi-s3-test-nginx
     image: nginx
     volumeMounts:
       - mountPath: /usr/share/nginx/html/s3
         name: webroot
  volumes:
   - name: webroot
     persistentVolumeClaim:
       claimName: bucket-csi-s3-manual-pvc
       readOnly: false

The STORAGECLASS/PVC/PV all create correctly, but the POD gets error "MountVolume.MountDevice failed for volume "manualbucket-with-path" : rpc error: code = Unknown desc = Timeout waiting for mount" and never completes.

[SeanHai]

I also encountered the same problem, anyone can help?

[xincan]

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: bucket-csi-s3-manual-pvc
spec:
storageClassName: ""
accessModes:
- ReadWriteMany
resources:
requests:
storage: 10Gi

storageClassName: "" 改为 storageClassName: "bucket-csi-s3"

[justbelka]

apiVersion: v1 kind: PersistentVolumeClaim metadata: name: bucket-csi-s3-manual-pvc spec: storageClassName: "" accessModes: - ReadWriteMany resources: requests: storage: 10Gi

storageClassName: "" 改为 storageClassName: "bucket-csi-s3"

I have the same error too. I changed it like you said, but it didn't work. Can you explain in more detail?
我也有同样的错误。 我按你说的改了，但没用。 你能更详细地解释一下吗?

[rhallier]

Hi,
I got the same issue whereas it worked correctly with a previous version of the chart ...

[vadimkim]

Same issue for me. I am also using pre-defined bucket name. POD is "creating" and never finishes.

[dzmitryastrouski]

Did you guys create Bucket first? Create Bucket manually or using the IaC approach. Only then will the tandem PVC -> PV -> Bucket work.

[scobit]

Any updates? I have the same issue.

[vadimkim]

@scobit I managed to solve the issue. The problem in my case was related to geesefs driver missing. I am using OKD + CoreOS on worker nodes and this OS has FUSE support, but no geesefs binary. You have to install it manually. In my case it is just downloading geesefs binary from Release page and put into /usr/local/bin. Then create symlink to it with name "mount.geesefs"
Keep in mind that default SELinux policy, if Enforced, might be blocking mount as well. Since it was a development cluster I switched off SELinux. For production system you need to write a policy.

[scobit]

@vadimkim Thank you, the problem was in SELinux .