Summary
This article can be used for DR to convert a multi-node cluster to a single node cluster on a VM.
Symptoms
You can only restore cluster nodes through BMR. This article describes the steps to perform a restore to a single node.
Resolution
It is possible to restore a multi-node cluster running either Windows 2003 or 2008 to a single VM. This assumes that the original cluster was configured so that it can run as a single node cluster. See technet.microsoft.com/en-us/library/cc770620.
Prior to running the BMR process
Ensure the DC is available for the new VM.
Configure the VM to match the original configuration of one of the physical nodes in the cluster. For example, if there are two NICs present on the physical machine, then add two NICs to the new VM.
In addition, configure the shared disks on the new VM as follows:
- Separate SCSI controller. If the main disks are on SCSI 0:x, then the shared disks should be configured on 1:x.
- Provision disks so that it can share with other VMs. This is done when new disks are added to the VM. The provisioning option is set in the Add Disk wizard. In vSphere 4.x, select disk provisioning with Support clustering features such as Fault Tolerance. In vSphere 5.x, select Thick provision Eager Zeroed.
After adding the disk, the ESX host automatically adds the SCSI Controller. Modify this controller to allow virtual disk sharing. This is especially important if the intention is to add new nodes to the cluster at a later stage.
Post BMR process
Start the new VM and log in to the machine with a Domain Admin account.
The following message may appear:
"The trust relationship between this workstation and the primary domain failed."
This can occur if the machine account has the incorrect role or its password does not match the domain database password. To resolve, place the server into a workgroup and then re-add to the domain.
If there are no authentication issues, the cluster service starts and all drives and services are immediately available.
If the drives are not available, then typically the shared disks are not configured correctly on the new VM or there is an authentication error for the cluster service. The following messages display:
- On Windows 2003:
"the drive is visible however when you access you receive message "<Drive> is not accessible. The device is not ready" - On Windows 2008:
the drive is not visible. Within Disk Management its shown as Reserved.
To resolve, remove the server from the domain and re-add. Or, if the shared disks were not provisioned using Thick Provisioning Eager Zeroed, re-run the BMR.