CephFS Unable to attach or mount volumes: unmounted volumes=[image-store]
In the pasted YAML for your StorageClass, you have:
reclaimPolicy: Deletea
Was that a paste issue? Regardless, this is likely what is causing your problem.
I just had this exact problem with some of my Ceph RBD volumes, and the reason for it was that I was using a StorageClass that had
reclaimPolicy: Delete
However, the cephcsi
driver was not configured to support it (and I don't think it actually supports it either).
Using a StorageClass
with
reclaimPolicy: Retain
fixed the issue.
To check this on your cluster, run the following:
$ kubectl get sc rook-cephfs -o yaml
And look for the line that starts with reclaimPolicy:
Then, look at the csidriver
your StorageClass is using. In your case it is rook-ceph.cephfs.csi.ceph.com
$ kubectl get csidriver rook-ceph.cephfs.csi.ceph.com -o yaml
And look for the entries under volumeLifecycleModes
apiVersion: storage.k8s.io/v1beta1kind: CSIDrivermetadata: creationTimestamp: "2020-11-16T22:18:55Z" name: rook-ceph.cephfs.csi.ceph.com resourceVersion: "29863971" selfLink: /apis/storage.k8s.io/v1beta1/csidrivers/rook-ceph.cephfs.csi.ceph.com uid: a9651d30-935d-4a7d-a7c9-53d5bc90c28cspec: attachRequired: true podInfoOnMount: false volumeLifecycleModes: - Persistent
If the only entry under volumeLifecycleModes
is Persistent
, then your driver is not configured to support reclaimPolicy: Delete
.
If instead you see
volumeLifecycleModes: - Persistent - Ephemeral
Then your driver should support reclaimPolicy: Delete