Reclaiming Space from a thin provisioned netword raid-5 volume

[psmiano]

And there are manual steps (CLI) in VMware to recover space on volumes:

http://vmfocus.com/2015/02/26/how-to-hp-storevirtual-lefthand-os-12-0-with-t10-unmap/

http://www.boche.net/blog/index.php/2012/06/28/storage-starting-thin-and-staying-thin-with-vaai-unmap/

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2014849

http://kb.vmware.com/selfservice/search.do?cmd=displayKC&docType=kc&docTypeID=DT_KB_1_1&externalId=2057513

[Patrick Neuner]

We did tests with Centos 6 and the discard feature. It does work with LH12 which are good news. 

Now the question is, why isn't it officially supported and about the risks. I don't really see them, but... still... 

[CalvinZito]

---

@Patrick Neuner wrote:

We did tests with Centos 6 and the discard feature. It does work with LH12 which are good news. 

Now the question is, why isn't it officially supported and about the risks. I don't really see them, but... still... 

---

The risks are unknown because the team didn't do testing with Centos 6.  The priority for the UNMAP feature was VMware and there hasn't been regressioning testing on any other OS that I'm aware of.  So using UNMAP on other OSes is certainly a risk. 

As I previously said, if you are looking for something to be supported that isn't today, you need to ask your HP Rep to submit an exception request. 

[oikjn]

hmmm...  got unmap working on NEW volumes, but it seems like existing ones are not wanting to cooperate.  Reboots and reconnects don't seem to work, but when I connect a new LUN to the same servers that new LUN will reclaim space.

I must say it reclaims space pretty quickly!  now I'll probably just migrate all the data around as I don't have enough for this to be pohibative and it will probably take less time than to try and figure out how to fix the problem, but I did follow the suggestions about doing a rescan as the help section suggests.

edit:  I was able to get the unmap to function by disconnecting the disk from the server, deleting the disk from computer manager and then reconnecting to it again.  All is working and its busy freeing up about at least 40% of my SAN storage!  Yay!

[a_o]

Any one have idea of to get Space reclamation to work with Win2008 R2?

As NTFS SCSI Unmap native support starts with Win2012, I figured there shoud be another way, albeit manually to get it to work with 2008R2.

I tried running Sdelete, which works with some other SAN's implementation of Unmap - i.e. NetAPP, Nimble, but the thin LUN just grew to it's thick size. (Sdelete -z basically zeroes all truly unused blocks in the Volume. I'used it for doing manual TRIM with OS direct attached SSDs.)

I also tried shrinking the volume from the  OS side, and Lefthand showed reduced consumed space to match  what the OS saw.  But once I increased the thinly provisioned LUN back to it's original size,  LH showed it as if it was fully provisioned.