3

I have been tasked with upgrading an existing RAID6 array with new disks. Currently we have a 6 disk setup with 1 hot-swap. This is inside a Dell R340 with a PERC H730 Mini controller. I have not been able to verify yet if hot-swap is enabled. The goal here is to replace the 200GB spinners with 800GB SSD's. My first plan was to pull one disk at a time and let it rebuild, but i understand that puts extra stress on the raid especially when it is in production. It was recommended to me by a couple other techs that I should add a big spinner to the last drive bay, and do a VHD copy over to the large disk. After that, I was told to replace the entire array with new disks and build fresh, reinstalling ESXi and copying over the VHD's

I am concerned with downtime, but can plan for it if needed, again, I'm looking for suggestion on best practices. Which method would you suggest? I know this question has been asked before, but i am looking for fresh information. Thanks!

1
  • 3
    If you replace the disks one at a time and have the array rebuild, you will end up with the same size array as when you started. You would able to create a second virtual disk out of the remaining space on the disks, though it would not be optimal. The best way to replace the drives is to backup, create the new RAID using the new disks, and restore to that.
    – Bert
    Oct 22, 2019 at 18:06

1 Answer 1

2

If you want to do this with no downtime, be SURE that you've got a good backup of the data beforehand.

You can actually accomplish this without too much hassle using the "replace member" feature/operation on a PERC 9 card. The feature is intended as a copy-back operation from a hotspare to a replacement drive, so that your configured hot spare drive "stays" as the hotspare when not in use. However, the operation can be started manually. The "replace member" operation would essentially mirror data from one of the drives in your RAID6 to a hot spare (which you should replace with a new larger capacity drive first). Since the data is mirrored, you're not throwing the entire RAID set into a degraded state each step of the way.

So the steps would look like:

  1. Unassign the current hot-spare, and swap it for a larger drive
  2. Assign the larger drive as the new hot-spare (this might not be necessary)
  3. Perform a "replace member" on one of the in-use smaller drives, selecting the new larger drive as the replacement
  4. Monitor the operation for completion
  5. Swap the drive you "replaced" in step 3 with a new larger drive, and repeat 1-5 until you're finished

Details on the "replace member" operation are on page 36 of the User's Guide

From there, you can actually expand the VD without downtime via "Online Capacity Expansion". Page 30 of the same user guide ("Reconfiguration of Virtual Disks" section) even mentions this "replace member" scenario specifically as a reason that you would have excess capacity to expand to.

Once all this is done, just perform a storage rescan from vSphere, and you should be able to expand the datastore size (though I think the VMFS filesystem might still be split into two pieces technically).

This whole process might not be the BEST way to do things, but it's certainly a Dell-supported method to get where you're going without the need for major downtime. But again, be 100% sure you've got a GOOD backup beforehand just to be on the safe side.

1
  • 1
    Right on!! Awesome explanation, thank you very much! Oct 25, 2019 at 22:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .