First, some background about my current setup (and I'll try to keep it short and sweet)...
I have 3 1TB hard drives in an ESXi 5.x host that I use in a RAID5 array. The RAID is actually set up and managed by a guest Windows Server 2008R2 VM on the host. When I created the datastores to build the array, I formatted each of the hard drives and added them inside the vSphere Client as VMFS datastores. When I added the drives to the guest VM's config, I set them to be thick-provisioned lazy-zeroed, and all 3 are the exact same size (930.2899996 GB). Each disk has one and only one VMDK, and that VMDK takes up most of the usable VMFS partition's usable sectors. Inside the guest, the virtual disks were formatted using Windows Disk Management as GPT dynamic non-boot disks and then set up into a RAID5 array.
A week or so ago during a VM backup/cleanup, I accidentally deleted a 2nd guest VM that also accessed those datastores (I was testing Windows Server 2012R2, unimpressed) without unlinking the datastores from the 2nd VM's config first, and ended up deleting the VMDK files from the disk. I am trying to recover the flat.vmdk files on at least 2 of these disks. Since this happened, I have spent many hours searching online about recovering VMDKs and running disk scans attempting to find the missing VMDK (as yet unsuccessfully). Nothing has been written to any of these disks, and the VMFS partition is still in-tact.
What I know...
Disk geometry, file names, logical sector size, and number of sectors of the missing VMDK files (retrieved from the log file of the VM's last successful boot)
Using the above info, I have a template created for a replacement VMDK descriptor file ready once I can extract the flat.vmdk images
What I have tried so far...
ESXi Shell & vmfs-fuse -> Browsing datastores shows the volumes are empty. vmfs-fuse mounts the volume, but no VMDK files can be found.
UFS Explorer -> Same basic idea. VMFS partitions are found easily and show as healthy, but I can't seem to find the files there, either.
Using UFS Explorer's hex view, I have spot-checked several areas that I know are inside the data area and verified the RAID5 parity checks out (confirmed), so it appears the data is still there and in-tact.
DiskInternals VMFS Recovery -> Attempted to scan the disk, gave up after 36 hrs of scanning on a single disk still hadn't reached 10%.
What I would like to do...
Find the starting location for the flat.vmdk file for each disk and do a binary image dump from that position to the end of the VMDK file (end sector calculated by start location + number of sectors for the VMDK from the log). Since all 3 disks are identical in the VMWare sense, this starting location should be the same for all 3 disks. I have a 3TB external HD on standby for the dump once I know where to start.
* The trouble I am having is actually locating the starting position in the hex view (I don't know what hex values I need to be searching for). From Sanbarrow's (continuum) site (VMDK-basics), he mentioned that he searches for the mbr as the file's header, but I haven't found that in my disks.
Final note 1: This is NOT a corporate or business system. It is my personal home server and I have legitimate licenses for all OS's running on the host.
Final note 2: Yes, I *DO* have a backup, but it is about 2 months old and missing about 300GB of files that were added since the last backup. I can recover a fair portion of the missing data (about 60GB) by going back to the source and re-copying/downloading the files, but there is still approx. 240GB that I can't re-copy (source no longer exists). My fault for not having more current backups, but worst-case is still not a TOTAL loss (approx 85% recovered from backup & original sources). All guest system disks VMDKs are stored on a separate SSD and have known good backups of the primary system disks on a separate drive).
Any assistance with this would be very much appreciated!