An unplanned permanent device loss (PDL) condition occurs when a storage device becomes permanently unavailable without being properly detached from the ESXi host.
The following items in the vSphere Web Client indicate that the device is in the PDL state:
The datastore deployed on the device is unavailable.
Operational state of the device changes to Lost Communication.
All paths are shown as Dead.
A warning about the device being permanently inaccessible appears in the VMkernel log file.
To recover from the unplanned PDL condition and remove the unavailable device from the host, you need to perform a number of tasks.
Power off and unregister all virtual machines that are running on the datastore affected by the PDL condition.
Unmount the datastore.
Perform a rescan on all ESXi hosts that had access to the device.
If the rescan is not successful and the host continues to list the device, some pending I/O or active references to the device might still exist. Check for virtual machines, templates, ISO images, raw device mappings, and so on that might still have an active reference to the device or datastore.