3PAR StoreServ Storage
cancel
Showing results for 
Search instead for 
Did you mean: 

HPE 3PAR unmapping free space

 
Highlighted
Occasional Visitor

HPE 3PAR unmapping free space

I`ve just taken over the management of a 3PAR SAN and I have no experience with the configuration.  As background I just migratate a 3.5Tb VM from one 3PAR SAN to another.  The migration completed succssfully but I then recieved an alert  "Total NL raw space usage at 82346G (above 95% of total 86664G)".  I only then realised that the LUN from VMWare (ESXi 6.0) are thin provisoned on the SAN side.  

I cleared around 2Tb of space from VMWare and then ran the "unmap" command against all the datastores on VMWare.  There after I ran CPG compact via the 3PAR GUI. 

FC free space went from 93% allocated to 92% allocated and NL went from 99% allocated to 98% allocated. 

Free space is being reported as below after the cleanup. 

cli% showspace
--Estimated(MB)---
RawFree UsableFree
1042432 521216

Any advise on how to proceed here?  Also what are the risks running at 90+ % allocation?

 

 

1 REPLY 1

Re: HPE 3PAR unmapping free space

There's a bunch of documentation at the HPE Information Library: www.hpe.com/info/storage/docs
It should come up already set to the 3PAR documentation area.

  • HPE 3PAR StoreServ Storage Concepts Guide - skim
  • HPE 3PAR Command Line Interface Administrator Guide - skim
  • HPE 3PAR StoreServ Storage and VMware vSphere 6.5 best practices - read
  • Adaptive Data Reduction Technologies - read

Risks running at 90+ % allocation? Short version - There's a limit to how much more data can be stored. If the system has virtual copies (snapshots) scheduled, if it runs out of free space snapshots could become "stale". Likewise, with the virtual volumes being thin provisioned, if a lot is added increasing a volume's allocation, that LUN could hang waiting for free space.

I wonder if you look again, if the numbers are any smaller. The command "showsys -space" will show a breakdown of all the storage capacity on the system. One key number is "uninitialized". If storage (1 GB chunklets) has been in use and released back to the system, the chunklets are put in the uninitialized queue to be scrubbed and moved to the initialized pool where they become available for use. If any chunklets are uninitialized, the system is already scrubbing them as fast as it can. Another command that looks at chunklets is "showpd -c" (show physical disks' chunklet usage). Amongst the columns are two; one for uninitialized "normal" chunklets, and one for uninitialized "spares". Again, if any are not zero, wait ten minutes and look again. Some numbers (look carefully) will be going down. And FC chunklets will re-initialize faster than NL chunklets.

As far as usable space, the command "showspace -cpg *" will display the estimated space for new and expanding logical disks in the "LDfree" column. Which does not mean you can create a virtual volume that big - that space is also needed for existing virtual volumes to allocate, along with any new snapshots.

If you still need help, HPE's Pointnext consulting can assist you further.


Note: While I am an HPE Employee, all of my comments (whether noted or not), are my own and are not any official representation of the company

Accept or Kudo