Description of Problem: oVirt csi driver should refuse to provision RWX and ROX PV Version-Release number of selected component (if applicable): 4.6.0-0.nightly-2020-09-24-162403 How Reproducible: Always Steps to Reproduce: 1. Create a PVC with storageclass created by oVirt csi operator # cat pvc.yaml kind: PersistentVolumeClaim apiVersion: v1 metadata: name: pvc-1 spec: storageClassName: ovirt-csi-sc accessModes: - ReadWriteMany resources: requests: storage: 1Gi 2. Create a deployment using this PVC 3. Set the deployment replicas=100 4. Check pods Actual Results: Some of Pods are in "ContainerCreating" status. Warning FailedAttachVolume 82s (x3 over 88s) attachdetach-controller AttachVolume.Attach failed for volume "pvc-94e0ea13-0fe7-4b46-9327-33495c01dde3" : rpc error: code = Unknown desc = Fault reason is "Operation Failed". Fault detail is "[Cannot attach Virtual Disk. Related operation is currently in progress. Please try again later.]". HTTP response code is "409". HTTP response message is "409 Conflict". Warning FailedAttachVolume 26s (x3 over 74s) attachdetach-controller AttachVolume.Attach failed for volume "pvc-94e0ea13-0fe7-4b46-9327-33495c01dde3" : rpc error: code = Unknown desc = Fault reason is "Operation Failed". Fault detail is "[Cannot attach Virtual Disk. The disk is not shareable and is already attached to a VM.]". HTTP response code is "409". HTTP response message is "409 Conflict". Expected Results: From the envent, seems oVirt volume should not support RWX and ROX PV, then the user create a PVC with `RWX` or `ROX` access mode, it should refuse to provision PV for it.
This looks like CSI driver issue. BTW, dynamic provisioning of ROX volumes does not make much sense, since the volume is empty and read-only empty volume is not really useful. It may get useful with cloning or snapshot restore (if oVirt supports attaching a read-only volume to multiple nodes). See here how Kubernetes PVC.Spec.AccessMode is translated into CSI capabilities: https://github.com/kubernetes-csi/external-provisioner/blob/8b0707649212d770624008edbd127f312121aff9/pkg/controller/controller.go#L395
due to capacity constraints we will be revisiting this bug in the upcoming sprint
need more time for this one
We can add a warning in the CSI driver logs for an unsupported accessmode
due to capacity constraints we will be revisiting this bug in the upcoming sprint Benny do you know when are you planning on taking this?
(In reply to Gal Zaidman from comment #6) > due to capacity constraints we will be revisiting this bug in the upcoming > sprint > > Benny do you know when are you planning on taking this? Not currently, depends on progress with my RHV work
Do we still need this now that we have block support?
(In reply to Gal Zaidman from comment #8) > Do we still need this now that we have block support? yes, we have to mark RWX volumes as "Shared" in RHV and
ocp: 4.10.0-0.nightly-2022-01-11-065245 1) create pvc with Access mode - ReadWriteMany 2) look on log of pod controller - excpected error appear: unsupported access mode MULTI_NODE_MULTI_WRITER, currently only RWO is supported pvc stuck on pending state but it's different bug as Vojech said
This bug has mistakenly been linked to two PRs: https://github.com/openshift/ovirt-csi-driver/pull/100 https://github.com/openshift/ovirt-csi-driver/pull/101 PR #100 was merged into 4.10, #101 was merged into 4.11. Cloning this issue to track the 4.11 bug.
The bug for the failing error code is 2043035
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: OpenShift Container Platform 4.10.3 security update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2022:0056