external-provisioner does not respect accessMode

[saad-ali]

external-provisioner hard codes access mode to:

	accessMode = &csi.VolumeCapability_AccessMode{
		Mode: csi.VolumeCapability_AccessMode_SINGLE_NODE_WRITER,
	}

It does not use the access mode specified on PV/PVC which means that the driver can not properly validate the accessMode.

Reported by @cduchesne!

[saad-ali]

In addition to passing the correct access mode to the CSI driver, after a volume is provisioned, external-provisioner should not just blindly set the access mode of the PV to the same access mode as the PVC.

Kubernetes allows a PVC to be fulfilled by a volume with looser access requirements. e.g. a PVC requesting RWO may be fulfilled by PV that is RWX. So PV access mode should be whatever the storage system specifies (not what the PVC says).

[saad-ali]

This is partially addressed by #133

[lpabon]

I'll take care of fixing this.

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[msau42]

/remove-lifecycle stale

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[msau42]

CSI CreateVolumeResponse does not have access mode. The assumption is that the volume provisioned must be able to support ALL the capabilities that were passed in the request.

 // The capabilities that the provisioned volume MUST have. SP MUST
  // provision a volume that will satisfy ALL of the capabilities
  // specified in this list. Otherwise SP MUST return the appropriate
  // gRPC error code.

So I think #133 is all we need for the fix.

/close

[k8s-ci-robot]

@msau42: Closing this issue.

In response to this:

CSI CreateVolumeResponse does not have access mode. The assumption is that the volume provisioned must be able to support ALL the capabilities that were passed in the request.

// The capabilities that the provisioned volume MUST have. SP MUST
 // provision a volume that will satisfy ALL of the capabilities
 // specified in this list. Otherwise SP MUST return the appropriate
 // gRPC error code.

So I think #133 is all we need for the fix.

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.