vSphere ESXi 7.x
vSphere ESXi 8.x
After the ESXi host was upgraded and rebooted, certain datastores were no longer visible because the underlying devices had been repurposed as Raw Device Mappings (RDMs) without properly unmounting the original VMFS volumes. As a result, although the datastore entries remained in the inventory, the VMFS partitions were no longer present on the devices. During the reboot, this caused the datastore mounts to fail, and the entries were subsequently removed from the inventory.
From the vmkernel.log, the datastores were already reporting as corrupted
2025-03-19T03:38:40.611Z cpu2:2099188)WARNING: FS3: 608: VMFS volume Datastore (Allocated to xxxxxx)/56491ec8-b0bc24bd-xxxx-xxxxxxxxxxxx on naa.6006016076a13xxxxxxxxxxxxxxxx:1 has been detected corrupted
The corruption warnings were triggered because the LUNs were reused for another purpose (as RDMs), while the VMFS volumes were still mounted. Upon checking, the affected devices were confirmed to be used as RDMs. The VMFS partition was overwritten and replaced with a Linux Native partition:
partedUtil getptbl /vmfs/devices/disks/naa.6006016076a13xxxxxxxxxxxxxxxxgpt668373 255 63 107374182401 2048 10737416191 EBD0A0A2B9E5443387C068B6B72699C7 linuxNative 0
Since the VMFS partition no longer exists, the volume fails to mount and the device shows up as Not Consumed in the UI.
In this scenario, recovery of the original VMFS datastore is not possible as the LUN has already been reformatted and is currently being used for other purposes (e.g., as an RDM). Since the datastore was placed in maintenance mode prior to the change, no data loss has occurred. No further action is required.