Page 1 of 1

Remove volumes not seeing space reclaimed

Posted: Fri Feb 28, 2014 12:26 pm
by evan631
I recently removed a few volumes, but I am not seeing the space as reclaimed. Do I have to wait for the 3par F400 to run some sort of maintenance?

Regarding the command compactcpg will this perform this for me? Do I need any switches? Any effect running this command during business hours?

Thanks for the help!

-Evan

Re: Remove volumes not seeing space reclaimed

Posted: Sun Mar 02, 2014 4:39 am
by c.uvaraja@gmail.com
hi,

Is this volumes were used by vmware datastore ? if this is the case then you may need to run vmkfstools tool to reclaim the blocks.

thanks
Uv

Re: Remove volumes not seeing space reclaimed

Posted: Thu Mar 06, 2014 4:55 pm
by Richard Siemers
Yes, after you delete volumes, it will return the space to the CPG... you should see more free space inside the CPG.... if you want that free space moved from the CPG to the system, compactcpg will do the trick.

Re: Remove volumes not seeing space reclaimed

Posted: Thu Nov 27, 2014 2:18 pm
by trarun0483
Hi
just try below dry run command and it will show you amount of space you will gain fater running compactcpg for required CPG

compactcpg -dr SQL_FC_SS6
Compacting CPG SQL_FC_SS6
Initial trim would reclaim 0 MB SA space, 0 MB SD space, and 0 MB USR space.
Would move 0 MB SA space, 0 MB SD space, and 0 MB USR space.
Final trim would reclaim 0 MB SA space, 0 MB SD space, and 0 MB USR space.

Re: Remove volumes not seeing space reclaimed

Posted: Fri Nov 28, 2014 8:50 am
by JohnMH
It depends on where you're looking for the space to be returned to.
When you delete volumes the chunklets first need to be cleaned before they are returned for use by the system, this is especially true for fpvv's as every chunklet is allocated upfront and so must be cleaned (zero's out) on deletion, you can see this by running the below.
CLI% showsys -space

If the uninitialized value is >0 it’s likely a large amount of data has been deleted and the blocks are in the process of being zero’d out before being returned to the system free pool. This process takes time and for non thin volumes every block allocated regardless of whether it was written needs to be zero’s out. There's no way to increase the speed of this process but it's only usually encountered during testing when large fully provisioned volumes are created, torn down and then recreated.

To view the cleaning process you can use :

CLI% showpdch –cng Specifies that information for chunklets that are being cleaned (reclaimed and zero’d after deletion) by the system are displayed, you can also use CLI% showpd –c

Once they're all clean a compactcpg which you should schedule will return these to the pool.