Can a disk be seperated/removed?

Question

Can disks be removed from an existing Raid Group and used to create a new Raid Group on CX array?

RRR · Accepted Answer

Once a disk is bound to a raid group, you can not remove that disk logically from the raid group.

I disagree: when disks 0_0_0 to 0_0_4 contain RAID Group 0 (with a few luns bound to it as well) and when you move a disk to a new location, it's no longer part of the RG is was a member of previously. Therefore it works like a charm when you pull for example disk 0_0_10 as well as disk 0_0_0 and place them in each other's slots, the "new" 0_0_10 is not a member of RAID Group 0 anymore and the "new" 0_0_0 is. The only thing you need to take into account is that in this example the new 0_0_0 should be able to contain the lun's that were on the old 0_0_0.

RRR · Answer

It will work, but what are you trying to accomplish ?

If you remove a disk from a working RG, the HS will kick in and the data will remain protected after the rebuild. The "old" disk can be plugged in another position and with other disks form a new RG again.

I did it before in a Clariion: swapping five 146GB disks with 73GB disks and create a new RG on the 5 146GB disk which are now in the back of the Clariion.

But I'd talk to your CE first before doing this.

kelleg · Answer

Once a disk is bound to a raid group, you can not remove that disk logically from the raid group.If you destroy the raid group, that will free up all the disks for reuse.regards,glen kelley

Kiran3 · Answer

if you want to free up disks and change layout, you can consider using lun migration to migrate the luns onto another free RG and then destroy this RG to use the disks as you wish to

sasamir1 · Answer

So the only option would be to copy data from this RG,unbind luns on RG > break RG, separate disks for creating new RG > bind luns > and restore the data as there is no other RG free to migrate the data...

Also can anyone tell how much time it would take to migrate data to new lun and also would it need a downtime if there would be performance concerns?

Kiran3 · Answer

migrate data priority can be changed (high low, asap...) but if your production is disk intensive then there could be a small performance hit...but then you can lower the migration priority.

RRR · Answer

High = 3MB/s, so 10GB per hour.

RRR · Answer

The first 10 disks were originally 146's and the last 5 were 73GB's.I tried drawing a picture for you, but somehow my html is not accepted and translated correctly.

RRR · Answer

New Clariion with 15 disksThe first 10 were 146GB and the last 5 were 73GB's.We meant to implement this as 5 x 73 and then the 10 146GB's. Unfortunatley the guy who implemented this, did not swap the disks in the beginning and therefore we had to swap them during production.And as I said, I'm not sure anymore whether or not there was a RG on these first 5 disks.And as I also said: I am going to test this in a few weeks when we're going to set up a new data center with new switches and Clariions...

dynamox · Answer

Rob ..i am confused. Where did disk 0_0_10 come from ? So you have a raid group (4+1R5) that spans disks 0_0_0 to 0_0_4, so what happens next ? You take disk 0_0_10 that's currently not used and you swap 0_0_0 with 0_0_10 ? Is 0_0_10 a higher capacity disk ?

Kiran3 · Answer

use [ code ] and [ / code ]

RRR · Answer

Thanks !Dynamox: do you see this ? This is what I meant !!Line 2 is what we got and line 3 is what we needed.

Kiran3 · Answer

Ooops, that didn't work eitherThe ASCII Painter is here to help you you probably meant this...0---1---2---3---4---5---6---7---8---9---10---11---12---13---14 146-146-146-146-146-146-146-146-146-146-73---73---73---73---73 73--73--73--73--73--146-146-146-146-146-146--146--146--146--146

RRR · Answer

Same here, but I guess my story is clear now.

Kiran3 · Answer

Thanks !np...I first thought that the reason for your table not to show up was M$ software that you used but I couldnt get simple HTML table to show up...not sure why!

CLARiiON

Was this post helpful?