Migrate a Cluster to a different SAN W2K / W2k3:
Before doing anything please have a full backup of all the nodes:
- Add new disk to SQL cluster group
- Take SQL resources offline
- Move all contents of the old drive into the new drive
- Change the drive letter of the old drive to an available letter
- Change the letter of the new drive to the letter of the old drive
- Bring SQL resources online
- Remove old drive.
- Detailed Process for changing disks (LUNs) or doing a SAN Migration:
In this scenario, we are going to use disk Z: as a drive we are going to replace.
The associated cluster resource for this disk is “Drive Z:”.
1. Create the new replacement LUN and present it to the operating system.
2. Once the operating system detects the disk, create a basic partition and format
it NTFS and give it a drive letter.
3. Copy the data from disk Z: to the new LUN.
4. Ensure all nodes in the cluster can see the new LUN.
5. In cluster administrator, add a new physical disk resource named “New Drive Z:”.
The new disk resource should be in the same resource group as the original Drive
6. Once “New Drive Z:” has been added, bring the resource online. Test failover
between all nodes.
7. Download the Cluster Recovery utility:
Cluster Server Recovery Utility (ClusterRecovery.exe)
8. Install and then execute the application. There are two options within the application:
1. Replace a physical disk
2. Reset checkpoints.
Choose replace a physical disk and click next.
9. You will then be presented with two drop-down boxes.
The first drop-down box you select the old drive you are about to decommission. In our scenario, you would choose “Drive Z:”.
10. In the second drop-down box, you would choose the new disk resource we created.
In our scenario you would choose “New Drive Z:” and click next.
This process changes all resources that depend upon the original Drive Z: resource to now depend upon the new Drive Z: resource.
It also renames the “New Drive Z:” to the name of the original resource, “Drive Z:”.
The original resource keeps the same name with “(lost)” appended after it.
One then can delete “Drive Z: (lost)” from cluster administrator and remove the LUN from the zone.
You would repeat this process for each disk you would like to replace.
Concerning the quorum resource – running \fixquorum and pointing to the new quorum disk will work.
Please note that the- new quorum disk should be presented to the nodes and in the cluster group before \fixquorum is run.
If you want to avoid rebooting (should be no big deal this is a cluster after all) doing a bus scan on Device manager -> disk drives -> scan for HW changes on ALL nodes 99.9% will eliminate the need for a reboot.
Additionally we can also use Dumpcfg to force the new signatures: