VM size is showing difference hosted on NFS and VMFS6 datastore

[Mayank Phirke]

Hi Team,

I have one VM of 18 TB (provisioned) but It is used only 8 TB so far, Tricky thing here is when VM resides on NFS3 datastore it shows 8 TB VM utilised but When i did storage migration from NFS to VMFS6 , Size become changed and now size is consuming on datastore is 16 TB.

Why does this difference ? How to reclaim that unused space and what is the mechanism behind that thing  

-------------------------------------------

[aweichenhain]

NFS can only handle thin provisioned vmdks.

where vmfs can use think provisioned Disks.

that might explain the difference

------------------------------
Axel
------------------------------

[QuikJean]

Are you sure you did not pick up “thick” while migrating ?




Unless otherwise stated above:

Compagnie IBM France
Siège Social : 17, avenue de l'Europe, 92275 Bois-Colombes Cedex
RCS Nanterre 552 118 465
Forme Sociale : S.A.S.
Capital Social : 664 614 175,50 €
SIRET : 552 118 465 03644 - Code NAF 6203Z

[janatlud]

Hello,

I don´t know why this happens, but it is common scenario. When you migrate VM to NFS datastore, it is Thin expected scenario. But, when you migrate the VM from NFS datastore, doesn´t matter, if you use thin disks. The disks are relocated in full size. Also the empty blocks are copied. It is also reason, why copying from NFS needs more time than copying to NFS. I´ve discovered the same issue. (We did it from vSAN and to vSAN. New server, where was only OS installed, but allocated disks were 2 TB. Copying to NFS in few minutes. Copying from NFS in several hours.

If the data are empty. The disk format is really THIN and is set automatic reclaim, is possible to run reclain command from OS perspective. This should shrink the disk again.

Here (on the bottom) are commands for Linux and Windows servers.

https://knowledge.broadcom.com/external/article/394562/trimunmap-not-working-on-vsan-cluster-ev.html

[jimfa]

Hello Mayank Phirke, 

Maybe when you did the storage migration, it went from thin provisioned to thick lazy zeroed, on the vmfs, 

You cloud try a storage migration again but flag is as thin provisioned, then the rethinning process will start, and hopefullly with will reduce in size, 

When you perform storage migrations of thin provisioned vm's, you have to be aware that the file type stays the same.

It also depends on what kind of storage array you have, if they are different brands, some force thick provisioned only by policy, that you need to change first.

These are some points to check, i hope this can help you with your issue.