Just to bump this thread.
I had a nightmare with metadata and datastore corruption warnings, vm's would not start, iso's would sometimes not mount, could not delete datastore, could not delete partitions. It was on an HPE Proliant G9 server, flattened the RAID array and rebuilt it, re-installed ESXi (6.5) to the USB stick and the datastore was still there! Performed a delete on the disks from the HP IP utility, reinstalled ESXi again and the datastore was still there! Booted from a Linux live usb (gparted) and it showed no partitions (perhaps cannot see vmfs).
So it would seem that rebuilding an array and deleting the disks on HPE servers does not get rid of the datastore.
Anyway, I traced the issue to an iso and a vmdk that would not delete. Tried every method mentioned here to no avail. What eventually worked was to move the files to a newly mounted NFS datastore. The files actually remained on the original datastore but a rm 'file.ext' deleted them.