AJK3

updated

17 years ago

A

AJK3

2 Intern

•

108 Posts

0

1402

December 8th, 2009 04:00

DRV Creation

I am trying to create DRV, but its failing saying that there is no enough space in the disk group to create the same. But i can see that there is 600+GB unconfigured capacity. All the devices on DMX are RAID-5.

Can any please help on how to get rid of this issue?

  • Quincy561

    1281 Posts

    585

    0

    Posted December 8th, 2009 04:00

    DRVs on DMX have to be unprotected devices, meaning all the capacity for the volume has to be contiguous on one physical disk.
  • AJK3

    2 Intern

    •

    108 Posts

    585

    0

    Posted December 8th, 2009 04:00

    Thanks, that was my question.
  • SKT2

    2 Intern

    •

    1340 Posts

    587

    0

    Posted December 8th, 2009 05:00

    where did u copy that from . I was looking for any info other than that..
  • Quincy561

    1281 Posts

    585

    0

    Posted December 8th, 2009 05:00

    It serves as protection (a mirror) as Optimizer is moving one of the members of a volume to a different physical drive.
  • mfvincent

    1 Rookie

    •

    62 Posts

    587

    0

    Posted December 8th, 2009 05:00

    DRV (Dynamic Reallocation Volume) A DRV is a non-addressable unit of storage used by the Optimizer to hold the

    user enterprise data while reconfiguration of the devices is executed."

    You can see the device, but you will be unable to use it.

    See:
    symdev show which will probably show the device status as "write Disabled"

  • SKT2

    2 Intern

    •

    1340 Posts

    585

    0

    Posted December 8th, 2009 05:00

    what does DRV exactly do??
  • Ed_R1

    53 Posts

    534

    0

    Posted December 9th, 2009 02:00

    No. When you're doing a meta expansion you need to use BCVs, which you've created for that express purpose.

    At least, assuming you're using something like:

    add dev F0AD:FAFF to meta ABCD, protect_data=true, bcv_meta_head=1356;

    You'll need to have created a (non RAID 5) BCV of the right geometry to hold a copy of the whole of the primary volume whilst the expansion is happening (e.g. if you can establish your volume to the BCV).


    But that's separate from a DRV, which is _only_ used for optimizer swaps.

    Anyway, the way you check how you're doing with free space is to do a symdisk list -gaps -v

    You might see something like:

      Gaps   (6):
        {
         Hyper   Hyper      Device  Device    Gap
         Before  After      Before  After     Size (MB)
         ------  -----      ------  -----     ---------
             13     14        2B41   051E           433
             75     76        1366   1159           730
             89     90        2C71   14BA            10
             92     93        06C7   2C80          2193
            101    102        2DA6   15B4          2921
            103      0        06CE   N/A           6330
        }

    Which is what I've got on a quick sample on my array.

    E.g. a bunch of gaps, in theory enough 'free space' for 10GB of hyper, but none of the gaps actually big enough.

    I've hacked together a bit of perl that does 'gap counting' if you're so inclined - it's a bit rough and ready though (I'm fairly sure the logic on figuring out what can be created is invalid for example) if you're interested?

  • SKT2

    2 Intern

    •

    1340 Posts

    587

    0

    Posted December 9th, 2009 02:00

    will they get used during a meta lun expand(striped meta; where data get redistributed across all componets)? Just a picked one possible situation...
  • Ed_R1

    53 Posts

    587

    0

    Posted December 9th, 2009 02:00

    Well, it is exactly what it says on the tin.

    A DRV is the 'holding area' when your symmetrix does a swap with optimizer, to prevent any disks going 'unprotected' whilst it's doing it.

    When it starts doing a swap, it attaches the DRV as an 'extra' mirror to the two hypers it's trying to swap.

    Data is replicated to the DRV, which temporarily takes on 'duty' as protection and performance.

    Hypers get swapped over, and data gets copied, and once the swap operation is complete, the DRVs get returned to the pool for re-use in the next swap operation.

    They need to be the same size (in cylinders) as the hyper that's being swapped, because it needs to fit the whole lot in. Even if it's a RAID 5, and it's only moving a single 'slice' of that RAID 5. (30GB made up as 7+1 Raid 5 is 8x 4-and-a-bit GB slices, but you still need the whole 30Gb available)

  • pappc

    1 Rookie

    •

    52 Posts

    407

    0

    Posted December 10th, 2009 09:00

    Bare in mind everything below is for DRVs with DMX-4 and lower Symmetrix models (things are slightly different with V-Max):

         A DRV device in the system is only used by the Optimizer feature for either Volume Swaps (swapping two or more device, in two device pairs), or Volume Migrations (move a volume to unused hard drive space).  Other features in the system such as Protected Meta Expansion do not use it.  (Meta Expansion uses BCVs as already mentioned).

         As mentioned before, a DRV is used for added protection to the data on the volume.  The reason for this is that during swapping or moving of data between physical locations in the array, the RAID protection of the volume is degraded until the syncing is complete.  As mentioned before, the DRV volume requires a mirror position on the swapping/moving device.  If a swapping/moving device has all of its available mirror positions taken up, for an example: a concurrent RDF RAID1 device (mmFF) or a RAID5 device with RDF and an established BCV (55FB), then a DRV is not available to be used and will be skipped for the process.  The only RAID protection that automatically skips the DRV requirement is RAID6, based on the fact that RAID6 is a dual-protection scheme already, so the extra protection of the DRV is not required.

    Hope this helps,


    PSE Chris