Provision a Hyper-V host or cluster from bare metal computers

Use this article to provision a Hyper-V host or cluster from bare metal computers with nothing installed on them, in the System Center - Virtual Machine Manager (VMM) fabric.

In addition to adding existing Windows servers to the fabric as Hyper-V hosts and clusters, VMM can discover physical bare-metal machines, automatically install an operating system, and provision them as Hyper-V server hosts and clusters.

Here's how you do this:

1. Verify prerequisites: Ensure you've all the prerequisites before you start.
2. Initial configuration: Set up the BIOS on the machine to support virtualization, set the BIOS boot order to boot from a Pre-Boot Execution Environment (PXE)-enabled network adapter as the first device, and configure the sign-in credentials and IP address settings for the BMC on each computer. You need to create DNS entries and Active Directory account for the machine names, and we recommend you allow time for DNS replication to occur.
3. Prepare the PXE server environment: Add the PXE server to VMM management (as described in Prerequisites: creating hosts, host clusters, Scale-Out File Server clusters from bare metal in VMM, and How to add a PXE server to VMM.)
4. Add resources to VMM library: Add resources that include a generalized virtual hard disk with an appropriate operating system (as listed in Prerequisites: creating hosts, host clusters, or Scale-Out File Server clusters from bare metal in VMM) to use as the base image and optional driver files to add to the operating system during installation.
5. Create profiles: In the library, create one or more physical computer profiles. These profiles include configuration settings, such as the location of the operating system image and hardware and operating system configuration settings.
6. Create Hyper-V host or cluster: You run different wizards depending on whether you want to set up a standalone host or cluster.

Before you start

Ensure the following prerequisites:

Add a PXE server to the VMM fabric

1. Select Fabric > Servers > Add > Add Resources > PXE Server.
2. In Computer name, specify the PXE server name.
3. Add the credentials for an account that has local administrator permissions on the PXE server. You can specify an existing Run As account (or select Create Run As Account to create a new one); manually enter user credentials in the format domain_name\user_name. Select Add.
4. In Jobs, verify that the job status is Completed and close the dialog. The job sets up the new PXE server, installs the VMM agent on the PXE server, imports a new Windows Preinstallation Environment (Windows PE) image, and adds the machine account for the PXE server to VMM.
5. Verify that the PXE server is added in Fabric > Servers > PXE Servers > Home > Show > Fabric Resources > PXE Servers. The agent status should be Responding.

Add and assign driver files

If you plan to assign custom drivers, the driver files must exist in the library. You can tag the drivers in the library so that you can later filter them by tag. After the files are added, when you configure a physical computer profile, you can specify the driver files. VMM installs the specified drivers when it installs the operating system on a physical computer.

In the physical computer profile, you can select to filter the drivers by tags, or you can select to filter drivers with matching Plug and Play (PnP) IDs on the physical computer. If you select to filter the drivers by tags, VMM determines the drivers to apply by matching the tags that you assign to the drivers in the library to the tags that you assign in the profile. If you select to filter drivers with matching PnP IDs, you don't need to assign custom tags.

1. Locate a driver package that you want to add to the library.
2. In the library share that is located on the library server that is associated with the group where you want to deploy the physical computers, create a folder to store the drivers and then copy the driver package to the folder.
3. We strongly recommend that you create a separate folder for each driver package, and that you don't mix resources in the driver folders. If you include other library resources, such as .iso images, .vhd files, or scripts with an .inf file name extension in the same folder, the VMM library server is unable to discover the resources. Also, when you delete an .inf driver package from the library, VMM deletes the entire folder where the driver .inf file resides.
4. In the VMM console, open the Library workspace. In the Library > Library Servers, expand the library server where the share is located, right-click the share, and then select Refresh. After the library refreshes, the folder that you created to store the drivers appears.
5. Now assign tags if required. In Library, expand the folder that you created to store the drivers in the previous procedure, and then select the folder that contains the driver package.
6. In the Physical Library Objects, right-click the driver .inf file, and then select Properties.
7. In the Driver File Name Properties > Custom tags, enter custom tags separated by a semicolon, or select Select to assign available tags or to create and assign new ones. If you select Select and then select New Tag, you can change the name of the tag after you select OK. For example, if you added a network adapter driver file, you could create a tag that is named ServerModel NetworkAdapterModel, where ServerModel is the server model and NetworkAdapterModel is the network adapter model.

Create a physical computer profile

1. Select Library > Home > Create > Physical Computer Profile.
2. In the New Physical Computer Profiles Wizard > Profile Description, type in a name and description and select VM host.
3. In OS Image > Virtual hard disk file > Browse, select the generalized virtual hard disk that you added to the library share. By default, if the disk is dynamic VMM converts it to a fixed disk during host deployment. We recommend that for production servers, you use a fixed disk to increase performance and help protect user data.
4. In Hardware Configuration > Management NIC, select the network adapter used to communicate with VMM and whether to use DHCP or a static address. If you want to use Consistent Device Naming (CDN) for the adapter or configure logical switches and ports, select Physical Properties. Select Add to add the adapter.
5. In Disk, specify the partitioning scheme for the first disk. You can use GPT if the physical computer profile is EFI. In Partition Information, select the volume label, whether to use all remaining free space or a specific size, and whether to designate the partition as the boot partition. You can also add a new disk or partition. During deployment, VMM copies the virtual hard disk file to the boot partition and automatically creates a system partition on the same disk.
6. In Driver filter, filter the driver filters to be applied to the operating system during host deployment. You can filter by Plug and Play ID or by specific tags. If you select to filter drivers with matching tags, ensure that you've added driver files to the library and assigned the corresponding tags.
7. In OS Configuration, specify the domain that the Hyper-V host or cluster should join, and specify local admin credentials and identity information. Add the product key for installation, and set the time zone. In GUIRunOnce, specify one or more commands to run when the user signs in to the Hyper-V host for the first time.
8. In Host Settings, specify the path of the host to store the files that are associated with virtual machines placed on the host. Don't specify drive C because it's not available for placement. If you don't specify a path, VMM placement determines the most suitable location.
9. In Summary, verify the settings. Wait until Jobs shows a status of completed, and then verify the profile in Library > Profiles > Physical Computer Profiles.

PCP post deployment settings

After you successfully create and deploy the PCP, you can configure additional settings such as RDMA, QoS, and SET using the PCP post deployment script.

Sample script

Here's the sample script to configure RDMA, SET, and QoS.

# Install data center bridging
Install-WindowsFeature Data-Center-Bridging

#Enable RDMA, assuming customer chosen switch name for storage as Storage1Switch and Storage2Switch
Enable-NetAdapterRDMA "Storage1Switch"
Enable-NetAdapterRDMA "Storage2Switch"

# set Qos Policy
New-NetQosPolicy "SMB" -NetDirectPortMatchCondition 445 -PriorityValue8021Action 3

# Enable net qos flow control
Enable-NetQosFlowControl  -Priority 3

# Disable net qos flow control other than 3
Disable-NetQosFlowControl  -Priority 0,1,2,4,5,6,7

# Enable net adapter qos on all adapters
Enable-NetAdapterQos  -InterfaceAlias "*"

# set qos traffic class
New-NetQosTrafficClass "SMB"  -Priority 3  -BandwidthPercentage 50  -Algorithm ETS

# Install windows feature
Install-WindowsFeature –Name Hyper-V

Install-WindowsFeature –Name RSAT-Hyper-V-Tools

# set net adapter property "encapsulated overhead"
NetAdapterAdvancedProperty -Name "*" -DisplayName "Encapsulated Overhead" -DisplayValue "160"

#disable ipv6
netsh int ipv6 isatap set state disabled

#Configure SET team mapping between virtual network adapter to physical network adapters. (Note: to get names of adapters, use command Get-NetAdapater. For team mapping use command
$physicalAdapters = Get-NetAdapter -Physical
$virtualStorageAdapter1 = Get-VMNetworkAdapter -ManagementOS | Where-Object {$_.Name -eq "Storage1Switch"}
$virtualStorageAdapter1 = Get-VMNetworkAdapter -ManagementOS | Where-Object {$_.Name -eq "Storage2Switch"}

Set-VMNetworkAdapterTeamMapping -ManagementOS -PhysicalNetAdapterName $physicalAdapters[0].Name -VMNetworkAdapterName $virtualStorageAdapter1.Name
Set-VMNetworkAdapterTeamMapping -ManagementOS -PhysicalNetAdapterName $physicalAdapters[1].Name -VMNetworkAdapterName $virtualStorageAdapter2.Name

#Set firewall rules.
[System.String[]]$Alias=@("vEthernet (WssdStorage2)", "vEthernet WssdStorage1)");
$Profile ='Any'
$Name="File and Printer Sharing"
$rules = Get-NetFirewallRule -DisplayGroup $Name

foreach ($rule in $rules)
     {
   		 $rule | Get-NetFirewallAddressFilter | Set-NetFirewallAddressFilter -  
         LocalAddress Any -RemoteAddress Any
     }

 Set-NetFirewallRule -DisplayGroup $Name -Enabled True -Profile $Profile – InterfaceAlias $Alias
 $Profile='Any'
 $Name=='FPS-LLMNR-In-UDP'
 Set-NetFirewallRule -Name $Name -Enabled True -Profile $Profile
 [System.String[]]$Alias=@("Storage2Switch", "Storage1Switch", "ManagementSwitch");
 $Profile ='Any'
 $Name="Windows Remote Management"
 Set-NetFirewallRule -DisplayGroup $Name -Enabled True -Profile $Profile –InterfaceAlias $Alias

 #Set assurance settings
 reg add HKLM\SYSTEM\CurrentControlSet\Control\DeviceGuard /v EnableVirtualizationBasedSecurity /t REG_DWORD /d 1 /f
 reg add HKLM\SYSTEM\CurrentControlSet\Control\DeviceGuard /v RequirePlatformSecurityFeatures /t REG_DWORD /d 2 /f
 reg add HKLM\SYSTEM\CurrentControlSet\Control\LSA /v LsaCfgFlags /t REG_DWORD /d 1 /f
 reg add HKLM\SYSTEM\CurrentControlSet\Control\LSA /v DisableRestrictedAdmin /t REG_DWORD /d 0 /f

Provision a Hyper-V host from bare metal

When you deploy a Hyper-V host from bare metal, VMM does the following:

1. Discovers the physical computer through out-of-band management.
2. Deploys an operating system image on the computer through the physical computer profile.
3. Enables the Hyper-V role on the computer.
4. Brings the computer under VMM management as a managed Hyper-V host.

Provision as follows:

1. Select Fabric > Servers > Home > Add > Add Resources > Hyper-V Hosts and Clusters.
2. In the Add Resource Wizard > Resource location, select Physical computers to be provisioned as virtual machine hosts.
3. In Credentials and Protocol select the Run As account with permissions to access the BMC. In the Protocol list, select the out-of-band management protocol that your BMCs use. If you want to use Data Center Management Interface (DCMI), select Intelligent Platform Management Interface (IPMI). Although DCMI 1.0 isn't listed, it's supported. Ensure the correct port is selected.
4. In Discovery Scope, specify the IP address scope that includes the IP addresses of the BMCs. You can enter a single IP address, an IP subnet, or an IP address range.
   • If you're provisioning a single computer, you can either specify a single IP address, or specify an IP address range that both starts and ends with the intended IP address. If you specify a single IP address, when you select Next, the computer is restarted.
   • If you specify an IP address range, when you select Next, information about the computer is displayed, and you can confirm that you specified the computer that you meant to.
5. If you specified a single IP address on the previous page, skip this step. Otherwise, the Target Resources page appears. Review the list of discovered BMCs (identified by IP addresses), and select the ones you want to provision as hosts. If you don't see all the BMCs that you expect, confirm that they are on a network accessible to the VMM server, and as needed, select Refresh.
6. In Provisioning Options, select a host group for new Hyper-V hosts. Select the physical computer profile you want to apply.
7. In Deployment Customization, review the list of computers again, and provide information for each computer that you want to include.
   • If you see a computer that you don't want to include, select the BMC (identified by IP address) and then select Remove.
   • To configure computers, select the BMC IP address.
   • Specify a unique computer name, without wildcard characters.
   • Select or clear Skip Active Directory for this computer name. The Active Directory check prevents deployment if the computer account already exists. This helps prevent deploying a computer with the same name as an existing computer. If you skip the Active Directory check, and there's an existing computer account in AD DS other than the Run As account that was specified in the physical computer profile, the deployment process fails to join the computer to the domain.
8. For the computer you're configuring, select a network adapter (on the left). You can modify the configuration, or fill in more information (...).
9. You can specify the MAC address of the management NIC (not the BMC) and static IP settings for this network adapter. If you specify an address, select a logical network and an IP subnet if applicable. If the selected IP subnet includes IP address pool, you can check Obtain an IP address corresponding to the selected subnet. Otherwise, type an IP address that's within the logical network or its subnet. If you select an IP subnet, ensure that it corresponds to the physical location where you're deploying the host and to the network that the adapter is connected to. Otherwise, the deployment may fail.
10. Configure the adapter settings for each network adapter.

14. Repeat the configuration for each BMC IP address in the list. When you've filled in information for all the computers you want to provision, select Next.
15. In Summary, confirm the settings, and then select Finish to deploy the new Hyper-V hosts and bring them under VMM management. Depending on your settings, the Jobs dialog might appear. Ensure that all the steps in the job have a status of Completed, and then close the dialog.
16. To confirm that the host was added, select Fabric > Servers > All Hosts > host group, and verify that the new Hyper-V host appears in the group.

Provision a Hyper-V cluster from bare metal

When you deploy a Hyper-V cluster from bare metal, VMM does the following:

1. Discovers the physical computers through out-of-band management.
2. Deploys an operating system image on the computers by using the selected physical computer profile.
3. Installs the failover clustering feature, and the Hyper-V role and MPIO feature.
4. Brings the provisioned cluster under VMM management.

Provision as follows:

1. Select Fabric > Servers > Add > Add Resources > Hyper-V Hosts and Clusters.

2. In General Configuration, specify a name for the host cluster. Choose a storage configuration if required:

   • For shared storage, select Storage connected to the cluster using shared SAS, FC, or iSCSI.
   • For Storage Spaces Direct, select Disk subsystem directly connected to individual nodes in the cluster.

3. In Resource Type > select Physical computers to be provisioned:

   • Specify the administrator Run As account to use for creating the cluster.
   • Select the physical computer profile (which provides the domain name and administrator Run As account for each node).
   • Next to the BMC Run As account box, select Browse, and select a Run As account that has permissions to access the BMC.
   • In the Out-of-band management protocol list, select the protocol that your BMCs use. If you want to use Data Center Management Interface (DCMI), select Intelligent Platform Management Interface (IPMI). Although DCMI 1.0 isn't listed, it's supported. Ensure that the correct port is selected.
   • If the Skip cluster validation option appears, and you don't need support from Microsoft for this cluster, you can skip validation.

4. In Discovery Scope, specify the IP address scope that includes the IP addresses of the BMCs. You can enter a single IP address, an IP subnet, or an IP address range. Deep discovery provides detailed information about a computer (for example, MAC addresses of network adapters), but restarts the computer, and requires additional time. You can allow or skip deep discovery.

5. If you specified a single IP address on the previous page, skip this step. Otherwise, the Target Resources page appears. Review the list of discovered BMCs (identified by IP addresses), and select the ones you want to include in the cluster.

6. If you don't see all the BMCs that you expect, confirm that they're on a network accessible to the VMM server, and as needed, select Refresh. Allow or skip deep discovery. Deep discovery provides detailed information about a computer (for example, MAC addresses of network adapters) but restarts the computer, and requires additional time. Then select Next.

7. In Deployment Customization, provide information for each computer that you want to include. If you see a computer that you don't want to include, select the BMC (identified by IP address) and then select Remove.

   • To configure computers, select the BMC IP address. Specify a unique computer name, without wildcard characters.
   • Select or clear Skip Active Directory for this computer name. The Active Directory check prevents deployment if the computer account already exists.

   Note

   If you skip the check and there's an existing computer account in AD other than the Run As account that was specified in the physical computer profile, the deployment process fails to join the computer to the domain.

   • For the computer you're configuring, select a network adapter. You can modify the configuration or fill in more information.
   • You can specify the MAC address of the management NIC (not the BMC) and static IP settings for this network adapter. If you specify an address, select a logical network and an IP subnet if applicable. If the selected IP subnet includes IP address pool, you can check Obtain an IP address corresponding to the selected subnet. Otherwise, type an IP address that is within the logical network or its subnet. If you select an IP subnet, ensure that it corresponds to the physical location where you're deploying the host and to the network that the adapter is connected to. Otherwise, deployment may fail.

8. Configure the network adapter settings for each network adapter.

9. Repeat the configuration for each BMC IP address in the list.
10. When you've filled in the needed information for all the computers you want to provision, select Next.
11. In Summary, confirm the settings, and then select Finish to deploy the new Hyper-V hosts and bring them under VMM management. Depending on your settings, the Jobs dialog might appear. Ensure that all the steps in the job have a status of Completed, and then close the dialog.
12. To confirm that the host was added, select Fabric > Servers > All Hosts > and locate and select the new host cluster. In the Hosts pane, in the Host Status column, verify that each node in the cluster is OK.