Creating a singular datastore from three spanned disks

The spanning is done by Windows, as I understand your situation (3 VMDK). Only Windows can put it all back together again. You can put all VHDX into the same datastore during Storage vMotion, but Windows will need to copy the data out of the spanned disk. Use robocopy.

I don't like to have really large volumes. If you have multiple shares, I recommend splitting them up among separate Windows disks so to keep each VHDK around 2 TB or less if possible.

VMware Converter May be able to help you

What there inside the guest disk Virtualization will not know. As you mentioned you need like to have a large disk (spanning disk)

The converter might copy all the data to one disk - that said its again a single .vmdk might not fit into the environment.

Since spanned disk considered as an only single disk with no parity, it is quite risky to have such disk and also it would take time if you wish to recover from backup.

It is a wise decision to divide your data into three (individual/drive) disks. Based on File size you may store large files and small files separately based on what storage you are using at the backend to improvise the performance too.

Since its file server, not a  database file. You would be able to easily split the data.

Isn’t this just a case of doing a full backup, redoing the underlying disk structure and replaying the backup

Converter will do what I need, but the file server would need to be down which is a dealbreaker. I'm now looking at the diskshadow/robocopy pairing which seems promising. What I don't see mentioned is once i make the shadowcopy and begin the robocopy, how do I ensure that files that have been added/changed between the beginning and end of the robocopy are copied also?? We're talking a 6tb file server that is used 24 hrs a day. Changes will happen.

Bill

Multiple runs of robocopy can get changed files copied over after the initial run.

I'm confused. My impression was that diskshadow creates a non-dynamic image of the drive. That said, re-running robocopy would yield the same results as the original run.

Ok, becoming clearer. But let's go back to the shadow copy. Does it change as the files on the file server change?

I have no idea how you would even attempt this with shadow copy. Any documentation on how you can use it to accomplish your goal?

THIS is a big part of my quandary. This is a file server used 24 hrs a day. Using Robocopy alone will miss open files. Thus I need diskshadow to stun open files so I can get them. All the while I need to go back and accommodate for added/changed/deleted files that came about during the robocopy process. And we haven't even gotten into preserving file permissions yet.

How bout this - diskshadow then robocopy. Then diskshadow and robocopy /xo??

Robocopy will only miss locked files. assuming the files that get locked keep changing, that isn't really too much of a problem. Robocopy can get them the next time. There is a time where you need to shut off write access to files for the final sync. Much like you need to stun a VM for vMotion.

I don't see how a shadow copy will help here, because you will want the very final version of all the files before you cut over. Shadow copy is designed to get you a copy of every file, but not the latest version.

robocopy /copyall will get NTFS permissions.

So plain ole robocopy, then maybe a robocopy /xo for fun then during maintenance window reboot server to sever file locks and do a final robocopy /xo and call it a day??

Similar problem at another site. https://community.spiceworks.com/topic/2183655-running-out-of-space-on-2tb-vhd-need-solution

I prefer
robocopy.exe [sourcedrive\parentdirectory] [destinationdrive] /e /mir /copyall /r:1 /w:1 /mt:20 /create

and then for all subsequent runs. For final run, you can change share permissions to read only or just kick people off and monitor open files. Whatever works for your environment.
robocopy.exe [sourcedrive\parentdirectory] [destinationdrive] /e /mir /copyall /r:1 /w:1 /mt:20