I just encountered the 16392 error myself and it really had me frustrated for a while. Turns out that the solution was to create the VMDK file in a datastore folder that was NOT the large 2TB+ location.
Here's more details of my setup for clarification and in case anyone else is looking for help. I'm still at the novice-intermediate skill level, so I apologize for anything that may be considered "poor practice".
Mounting a 2TB+ RAID drive or virtual disk as a raw device mapping (RDM) using ESXi 5:
Eight 3TB disks in a RAID 5 configuration using a LSI Logic controller. From within the RAID Controller 2 virtual disks were created. The first was about 1.5 TB and was for the purpose of storing all of the VMs. The remaining space, approx 17TB, was to be used for as file server storage.
- From within vSphere, mounted the 17TB disk. (Select server -> Configuration -> Storage -> Devices)
- Opened up SSH and logged in using PuTTY as root.
- Did the following as root:
Used a combination of the next two commands to ensure I had the proper vm1.REALLY_LONG_HEX_OF_DISK value. Not sure if these steps work with the naa.BLAHBLAH identifier.
esxcfg-mpath -l
ls /dev/disks -l
cd /vmfs/volumes/VMs (that's the friendly name of my 1.5TB datastore)
vmkfstools -z /vmfs/devices/disks/vm1.REALLY_LONG_HEX_OF_DISK data_storage.vmdk -a lsilogic - Go back to vSphere and add a hard disk to the VM with Use an existing virtual disk. -> Browse to your new VMDK on the OTHER datastore.
- TIP: Use Virtual Device Node SCSI (1:0), not (0:1)! Not sure why, but I saw this on another site while looking around and I figured it wouldn't hurt.
- Once done, I saw a new Mapped Raw LUN and LSI Logic Parallet device on my VM.
- Boot your VM and follow your OSes instructions for mounting a new disk
- Sit back, have a root beer, and bask in your awesomeness.
![]()
