Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

We currently have a server with 8 x HDD slots. It is a HP DL380G5 with a P400 controller.

2 x HDD are in a RAID 1+0 config and this hosts the OS.

6 x HDD are in a RAID 5 config and holds an Oracle DB.

Basically the RAID 5 volume is running out of space and we would like to swap all 6 with higher capacity disks.

Excuse my ignorance as I am pretty new to this... I believe we will need to backup the data, delete the RAID volume, insert the new disks, recreate the volume, and restore the data.

2 questions:

  1. Do we need to worry about the OS partition or is it completely independent so we can simply take out the 6 and insert 6 new disks and get the controller to recognize the 6 new disks and form a new RAID 5 volume? We should not need to reinstall OS or Oracle correct?
  2. Since we are going to restore the data on the volume from another source (our vendor will take care of this) but we would like to keep the existing data on the 6 disks just in case we run into issues and want to fall back, is this possible?

Thanks in advance.

share|improve this question

1 Answer 1

up vote 2 down vote accepted

First. The OS partition is completely independent. You should be able to mess with the RAID-5 array without affecting the OS. However. I presume from your question the plan is something like

  • replace one disk of the RAID-5 array with a larger disk and let it resync
  • repeat for the other drives in the array

This is technically possible, however you won't see any increase in available space, because at each stage the array is regenerating the missing, small-sized partition onto the larger disk, and when you are finished you will still only have the small-sized partition written across the larger disks.

Your best bet is to

  • stop Oracle
  • back up the whole system
  • remove the small disks and un-define the existing RAID-5 array
  • install the larger disks and redefine the RAID-5 array
  • define the new, larger partition(s)
  • restore the RAID-5 data from the backup

Second, since your controller is a HW controller, it might (might!) be possible to keep the old drives around to use as a backup/archive; however I wouldn't count on it. Because the disk sizes are changing you almost certainly have to destroy and redefine the array to see the larger disk sizes; to go back to the smaller disks you'd have to do it again, and there is a pretty good chance that the controller won't trust what is written to the disks and will effectively initialize the array (ie wipe the disks).

(Incidentally, your OS partition is probably a RAID-1 partition, not 1+0.)

share|improve this answer
    
@David - Thanks for your feedback. Your reply was exactly what I wanted to confirm and it feels good when your question makes sense to someone :P. What we were intending to do was not the "replace one disk and resync", instead, we intend to perform the steps you recommended. I read from a post relating to DELL that we could remove a disk and insert a higher capacity one and let it rebuild but as you mentioned, we would not be effectively increasing space and I just wanted confirmation of this. Looks like the old disks is not a backup option. As for OS partition, it says 1+0 in HP ACU? –  molecule Apr 27 '10 at 4:01
    
That's normal for SmartArray controllers. With a two-disk set, it's just doing RAID 1 even though the ACU says RAID 1+0 (the latter requiring a minimum of 4 disks). There's a short thread about it on HP's IT Resource Center forums: forums13.itrc.hp.com/service/forums/… –  James Sneeringer Apr 27 '10 at 4:24
    
@James: Thanks James. Based on your experience, how much time do you think it would take to rebuild a RAID volume with 6 x 146GB disks? I am just talking about the rebuild and not touching on restoring the data. I have not done this before but I believe deleting the RAID is simply a matter of clicking the Delete option on HP ACU? –  molecule Apr 28 '10 at 4:06
    
I've managed DL380G5 servers with that exact configuration, but I don't recall ever having to actually rebuild parity due to a disk failure. I've had to swap out SAS disks before, but as luck would have it, it was never on a RAID5 volume. I would guess it would take a couple hours at a minimum, more if the volume has a high I/O load. As for deleting the RAID set, but I believe you are correct that it's just a matter of choosing "delete" in the ACU. –  James Sneeringer Apr 28 '10 at 15:48
1  
As far as I know, it does not low-level format the disks themselves, it just wipes the RAID configuration. If you're following David's suggestion, you will need to do this, though it shouldn't matter whether you do it before or after you swap the disks; if you swap the disks first, the controller will just complain that it couldn't find the old disks, and then you'll just proceed into the ACU to delete and recreate. –  James Sneeringer Apr 28 '10 at 18:20

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.