HPE Storage Users Group
https://3parug.com/

Reclaiming unused space from ESX datastores
https://3parug.com/viewtopic.php?f=18&t=2087
Page 1 of 1

Author:  InteraX [ Fri Jun 24, 2016 10:42 am ]
Post subject:  Reclaiming unused space from ESX datastores

Hi guys,

In the 'HP 3PAR StoreServ Storage and VMware vSphere 6 best practices' document, HP state 'Thin persistence can reclaim space in chunks of 128MB of contiguous unused or zero-filled space on the LD'.

In the VMware doc 'Using esxcli in vSphere 5.5 and 6.0 to reclaim VMFS deleted blocks on thin-provisioned LUNs (2057513)' VMware say 'The default value of 200 for the -n number or --reclaim-unit=number argument is appropriate in most environments, but some array vendors may suggest a larger or smaller value depending on how the array handles the SCSI UNMAP command'.

Are HP saying that the reclaim-unit argument should be 128 or is the default value of 200 still adequate?

Thanks.

Author:  nabarry [ Wed Jun 29, 2016 2:01 pm ]
Post subject:  Re: Reclaiming unused space from ESX datastores

My understanding is:
Vmware sends the unmap command on 200MB chunks. This is a convenient number, and nothing in HPE's documentation countermands this as far as I can tell. This space is returned to the volume directly, leading to the disparity between reserved and used space.

In the background, completely outside of your control, the array may return logical disks to the CPG. This space is then available for any volume attached to the CPG, and can be freed to the completely free pool by running a compactcpg. Unfortunately, it is entirely possible for terabytes of space to stay attached in the reserved but unused state for at least 6 months (still counting on mine!), and there's nothing you can do about it aside from migrating off that volume and deleting it.

Author:  InteraX [ Thu Jun 30, 2016 4:34 am ]
Post subject:  Re: Reclaiming unused space from ESX datastores

I don't think it's acceptable that we should have to wait more than a few weeks for unused space to be released back to the system. HP are selling the zero detect feature with the "start thin, get thin stay thin" sales pitch.

I've raised this via support. I would expect to see at least 100GB per day to be released back to the system. It's just been raised to L2, so will have to see what they say.

Author:  MammaGutt [ Thu Jun 30, 2016 6:25 am ]
Post subject:  Re: Reclaiming unused space from ESX datastores

What type of 3PAR you got?

My experience is that this seems to be somewhat resource based. On 7200s, things take time. Some improvement on the 7200c, but generally 7400 and above tend to reclaim faster. 3PAR OS 3.1.2 was also painfully slow no matter what, 3.2.1 seems better. No experience with 3.1.3 or 3.2.2.

Author:  InteraX [ Thu Jun 30, 2016 6:47 am ]
Post subject:  Re: Reclaiming unused space from ESX datastores

They're 7200s running 3.2.1 MU2.

Hopefully support will be able to advise us on how this space can be reclaimed quicker.

Author:  nabarry [ Thu Jun 30, 2016 8:34 am ]
Post subject:  Re: Reclaiming unused space from ESX datastores

I have a 4-node 7400 which was on 3.2.1 MU2, and is now on 3.2.2 EMU2. I logged the same ticket about 6 months ago now. There's no way to accelerate it, it's done based on how busy the LUN is. The only way to view its progress is from CLI to run showeventlog -oneline -debug -min (how far back you want to go) -msg reclaim, and if you're on 3.2.2 you'll want to add -type notification as in 3.2.2 they intro'd a log that just spams that there is space to be reclaimed.

That will show you which volumes are reclaiming space, SD2 is snapshot space, SD is actual user space. I have a volume that has had over a TB to be reclaimed for 6 months now. I've also got some volumes that have reclaimed a couple terabytes in the course of a few weeks to a month. 3.2.2 EMU2 doesn't appear to be accelerating the reclaim process unfortunately.

Author:  sjm [ Thu Jul 14, 2016 8:41 pm ]
Post subject:  Re: Reclaiming unused space from ESX datastores

My 2 x7400 sans are reclaiming reasonable amounts every 24 hour period.

But I do run compact cpg's tasks every night except our full backup window nights

Author:  nabarry [ Wed Dec 02, 2020 9:31 am ]
Post subject:  Re: Reclaiming unused space from ESX datastores

Primera has added a compactvv command to condense the VV immediately. I've been asking for this for 5 years... and I'm currently bugging product management to backport to 3PAR.

Author:  Marc.mvh.vanhoof [ Thu Dec 03, 2020 1:12 pm ]
Post subject:  Re: Reclaiming unused space from ESX datastores

More info available on compactvv with Primera? Lets hope one can do compactvv when remote replication on vv is active :D

Page 1 of 1 All times are UTC - 5 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/