Unable to claim disk for vSAN OSA

Products

• VMware vSAN

Issue/Introduction

• Users are unable to claim specific SSDs on host nodes while attempting to build a vSAN Original Storage Architecture (OSA) cluster on vSAN 8.0 U3.
• Attempts to erase disk partitions using the ESXi vSphere Client (GUI) fail.
• Executing partedUtil commands in the ESXi CLI to clear the disk fails, returning errors such as "Read-only file system" or "No such file or directory".
• The system may display "Permanent Disk Failure" or "Cannot change host configuration" errors during the disk erasure process.
• The host recognizes the drive as containing an existing Operating System (e.g., legacy Nutanix CVM data), even though the drive is not actively being used by ESXi for boot, scratch, or core dumps.

Environment

VMware vSAN 8.0 OSA

Cause

This issue occurs because the disks contain persistent metadata and partitions from a previous third-party installation (such as a Nutanix environment).

The ESXi VMkernel frequently flags these "Foreign" or "OS" partition tables as active system disks. To protect what it perceives as an active OS, the VMkernel places a strict lock on the partition table. Because the original third-party management tools are no longer present to release the disks, native ESXi tools (like partedUtil) are blocked from overwriting the GUID Partition Table (GPT) or modifying the drive.

Resolution

To resolve this issue, you must bypass the ESXi software layer and its kernel locks to completely clear the legacy metadata from the disk.

1. Boot into a live utility: Create a bootable USB with a third-party Linux partition manager, such as GParted Live, and boot the affected ESXi host from the USB.
2. Locate the disk: Within the partition utility, identify the affected SSD(s).
3. Clear the partitions: Delete all existing partitions on the affected drive.
4. Create a new partition table: Initialize the drive by creating a new, blank partition table (GPT). Leave the overall disk space strictly unallocated.
5. Reboot the host: Reboot the host back into ESXi. (Note: If the boot drive was affected or wiped during this process, reinstall ESXi). 6. Claim the disk: Refresh the storage controllers in the vSphere Client. ESXi will now recognize the SSD as a "raw" disk, allowing you to successfully initialize and claim it for the vSAN OSA cluster.