[manila-csi-plugin] Cannot extend the same volume twice

[m-k8s]

Is this a BUG REPORT or FEATURE REQUEST?:

/kind bug

What happened:

When I extend the volume a second time I get this error message:

volume pvc-8f8e61c9-18b5-4706-9161-8346ddcc55bc is already being processed

What you expected to happen:

Resize volume succeeded

How to reproduce it:

1. Create a persistentVolumeClaim:

cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: client-pvc
  namespace: test-csi-nfs
spec:
  storageClassName: csi-manila-nfs
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: 1Gi
EOF

2. Extend the volume for the first time by switching it to 2Gi, the expansion will work correctly

kubectl -n test-csi-nfs patch pvc client-pvc -p '{"spec":{"resources":{"requests":{"storage":"2Gi"}}}}'

3. Extend the volume again by switching it to 3Gi:

kubectl -n test-csi-nfs patch pvc client-pvc -p '{"spec":{"resources":{"requests":{"storage":"3Gi"}}}}'

There is an error :

kubectl -n test-csi-nfs describe pvc client-pvc
...
Events
...
Warning  VolumeResizeFailed      0s (x2 over 2s)   external-resizer nfs.manila.csi.openstack.org    resize volume "pvc-c7a34569-7a3e-4b49-acf3-520f7e0d8651" by resizer "nfs.manila.csi.openstack.org" failed: rpc error: code = Aborted desc = volume pvc-c7a34569-7a3e-4b49-acf3-520f7e0d8651 is already being processed

Anything else we need to know?:

At line 451 of the controllerserver.go file, I think we should use the value of req.GetName() as the key to remove the volume from the pendingVolumes list instead of the value of share.Name

So when we extended the volume the first time, this volume always stayed in the pendingVolumes list while the volume is ready

https://github.com/kubernetes/cloud-provider-openstack/blob/master/pkg/csi/manila/controllerserver.go#L448-L451

Environment:

• openstack-cloud-controller-manager(or other related binary) version: 1.22.0
• OpenStack version: train

[m-k8s]

/assign

[gouthampacha]

Thanks for reporting this and for working on a fix, @alexisries - we'll get the patch reviewed.