Clustered VMDK support unavailable on VMFS 6 with PowerFlex scini adapter

Products

• VMware vCenter Server

Issue/Introduction

When migrating clustered virtual machines (VMs) to VMFS 6 on ESXi 8.x, the Windows Server Failover Clustering (WSFC) software cannot locate or bring the shared disks online. The option to Enable Clustered VMDK Support is missing or greyed out in the vSphere client. This occurs specifically when using the PowerFlex Storage Data Client (scini) virtual adapter.

Symptoms

• After migrating clustered Windows VMs from vSAN to VMFS 6 on ESXi 8.0 Update 3, the Windows Server Failover Clustering (WSFC) software cannot locate or bring the Quorum disk online.
• The option to Enable Clustered VMDK Support is missing or greyed out under the Datastore Capabilities in the vSphere Client.
• The storage adapter is identified as scini (PowerFlex Storage Data Client), which presents as a Virtual Fiber Channel (FC) adapter (e.g., vmhba64 scini link-up fc... FC(virtual)).
• Attempts to share standard virtual disks (.vmdk) between VMs fail despite proper SCSI controller configurations.

Environment

• vCenter Server: 8.0.3 (Build 24322831)
• ESXi: 8.0.3 (Build 24784735)
• Storage Adapter: scini (Dell PowerFlex Storage Data Client)
• Storage Protocol: Virtual FC (presented by SDC)
• File System: VMFS 6

Cause

The "Clustered VMDK" feature, which allows sharing of standard virtual disks (.vmdk) on VMFS 6, is not supported when using the PowerFlex SDC (scini) driver.

Per Broadcom KB 312201, Clustered VMDK support on PowerFlex is exclusively limited to NVMe/TCP configurations (PowerFlex 5.0/VCF 9.0 or PowerFlex 4.x on ESXi 8). Because the scini virtual adapter does not support the required primitives, standard .vmdk files cannot be used as shared cluster disks in this specific configuration.

Resolution

To support SCSI3-Persistent Reservations (SCSI3-PR) required by WSFC on PowerFlex SDC, you must use Physical Compatibility Mode Raw Device Mappings (RDMs) instead of virtual disks. 

Step 1: Configuration on the Primary Node

1. Right-click the VM in the vSphere Client and select Edit Settings.
2. Click Add New Device and select RDM Disk.
3. Select the target LUN intended for the cluster (presented via the scini adapter) and click OK.
4. Choose a datastore to store the RDM pointer file (.vmdk). Broadcom recommends selecting Store with the virtual machine or browsing to a shared datastore accessible by all nodes.
5. In the Compatibility Mode dropdown, you must select Physical.
6. Set Sharing to No sharing (sharing is managed at the SCSI controller level).
7. Assign the disk to a dedicated SCSI controller (e.g., SCSI 1:0). Note the exact ID used (e.g., 1:0).
8. Ensure the dedicated SCSI controller has SCSI Bus Sharing set to Physical.

Step 2: Configuration on Additional (Secondary) Nodes

1. Right-click the secondary VM and select Edit Settings.
2. Click Add New Device, select Existing Disk, and browse to the datastore/folder where you stored the RDM pointer file created in Step 1.
3. Manually select the Virtual Device Node that matches the primary node (e.g., if the primary is on SCSI 1:0, the secondary must also be on SCSI 1:0).
4. Ensure the SCSI controller on this VM also has SCSI Bus Sharing set to Physical.

Validation

1. Verify the disk mode is Independent - Persistent.
2. Run the Windows Cluster Validation Wizard to confirm SCSI-3 PR compliance.
3. Apply DRS Anti-Affinity rules to ensure VMs reside on separate ESXi hosts.

Additional Information

• For more information on supported PowerFlex configuration, see Dell PowerFlex with ESXi 8.0 and 9.0.
• Review the Microsoft Windows Server Failover Clustering Guidelines for general best practices. 
• To receive updates on this article, see How to Subscribe to Broadcom Articles. 
• To speak with a customer representative or Support Engineer, see Contact Support. Scroll to the bottom of the page and click on your respective region.