3-Node Hyper-V Failover Cluster
on Windows Server 2022

Windows Server Failover Clustering (WSFC) with Hyper-V gives you enterprise-grade VM high availability — if one node goes down, VMs automatically restart on surviving nodes. This guide covers the exact setup used in the homelab: three AZW SER mini PCs running Windows Server 2022 with a dedicated storage VLAN and CSV shared storage.

Hardware Used

1. 1

   Download & Create Install Media

   Download the Windows Server 2022 ISO from the Microsoft Evaluation Center (free 180-day eval). Write to USB with Rufus — use GPT partition scheme for UEFI systems (most modern hardware).

   If the installer fails to detect your NVMe drive, the AZW SER may need a driver injected into the ISO. Use DISM or the Rufus "Add drivers" option to inject the NVMe driver.
2. 2

   Install on All 3 Nodes

   Select Windows Server 2022 Standard (Desktop Experience). Perform a clean install. Assign static IPs before proceeding:

   # PowerShell — set static IP New-NetIPAddress -InterfaceAlias "Ethernet" -IPAddress 10.10.50.X -PrefixLength 24 -DefaultGateway 10.10.50.1 Set-DnsClientServerAddress -InterfaceAlias "Ethernet" -ServerAddresses 10.10.50.10,10.10.50.11
3. 3

   Set Hostnames & Activate

   Rename-Computer -NewName "HV-NODE01" -Restart # Repeat: HV-NODE02, HV-NODE03

Failover Clustering requires all nodes to be domain-joined. Install AD DS on the first node to make it your domain controller, then join the others.

1. 1

   Install AD DS on Node 1

   Install-WindowsFeature -Name AD-Domain-Services -IncludeManagementTools Install-ADDSForest -DomainName "homelab.local" -InstallDns
2. 2

   Join Remaining Nodes to Domain

   On NODE02 and NODE03, point DNS to NODE01's IP, then join the domain:

   Add-Computer -DomainName "homelab.local" -Credential (Get-Credential) -Restart
3. 3

   Create a Cluster Service Account

   In Active Directory Users and Computers, create a dedicated service account (e.g. svc-cluster) with a strong password and no expiry. This account will be used for cluster operations.

1. 1

   Install Hyper-V Role

   Run on each node:

   Install-WindowsFeature -Name Hyper-V -IncludeManagementTools -Restart
2. 2

   Create the External Virtual Switch

   In Hyper-V Manager → Virtual Switch Manager → New → External. Bind to the physical NIC that connects to your LAN. Name it consistently across all nodes (e.g. External-LAN) — the name must match for live migration to work.

Separate your storage traffic from VM traffic using VLANs. This prevents a busy VM from saturating the storage link during CSV I/O.

1. 1

   Create a Dedicated Storage NIC Team or VLAN

   On your managed switch, create a storage VLAN (e.g. VLAN 20). On each node, add a second NIC or create a VLAN adapter:

   # Add VLAN-tagged adapter for storage network Add-VMNetworkAdapter -ManagementOS -Name "Storage" -SwitchName "External-LAN" Set-VMNetworkAdapterVlan -ManagementOS -VMNetworkAdapterName "Storage" -Access -VlanId 20 New-NetIPAddress -InterfaceAlias "vEthernet (Storage)" -IPAddress 10.10.20.X -PrefixLength 24
2. 2

   Configure Live Migration Network

   In Failover Cluster Manager → Networks, set the storage network to Allow cluster network communication only and the LAN to Allow all. Prioritize the storage network for live migration traffic.

1. 1

   Install Failover Clustering Feature

   # Run on ALL nodes Install-WindowsFeature -Name Failover-Clustering -IncludeManagementTools
2. 2

   Run Cluster Validation

   Always validate before creating — it identifies networking, storage, or AD issues before they become cluster problems:

   Test-Cluster -Node HV-NODE01,HV-NODE02,HV-NODE03

   Review the HTML report it generates. Warnings about storage are OK if you're using SMB storage (not iSCSI). Errors must be resolved first.

3. 3

   Create the Cluster

   New-Cluster -Name "HVCLUSTER01" -Node HV-NODE01,HV-NODE02,HV-NODE03 -StaticAddress 10.10.50.50 -NoStorage

   The -NoStorage flag skips auto-discovery — you'll add storage as a CSV manually in the next step. Give the cluster its own static IP (the Cluster Name Object, or CNO) — this is how Failover Cluster Manager connects to it.

4. 4

   Configure Quorum

   With 3 nodes, dynamic quorum works fine. For extra resilience, add a file share witness on a separate machine (e.g., your NAS):

   Set-ClusterQuorum -FileShareWitness "\\NAS01\quorum"

CSV allows all nodes to access the same storage volume simultaneously — required for live VM migration without downtime.

1. 1

   Add Storage to the Cluster

   If using an iSCSI target, connect each node to it first via iSCSI Initiator. Then in Failover Cluster Manager → Storage → Disks → Add Disk. The shared disk will appear if all nodes can see it.

   For SMB/NFS-based storage (e.g., NAS), use Scale-Out File Server or mount the SMB path directly as a cluster disk via Add-ClusterSharedVolume.

2. 2

   Convert to CSV

   # In Failover Cluster Manager: right-click disk → Add to Cluster Shared Volumes # Or PowerShell: Add-ClusterSharedVolume -Name "Cluster Disk 1"

   CSV volumes appear on all nodes at C:\ClusterStorage\Volume1\. Store VM files here.

1. 1

   Create a Test VM on CSV Storage

   In Hyper-V Manager (connected to the cluster), create a new VM and store its VHDX on C:\ClusterStorage\Volume1\. Boot the VM and confirm it's running.

2. 2

   Live Migrate via Failover Cluster Manager

   In Failover Cluster Manager → Roles, right-click the running VM → Move → Live Migration → Select Node. The VM moves to the target node while running — no downtime.

   # Or PowerShell: Move-ClusterVirtualMachineRole -Name "TestVM" -MigrationType Live -Node HV-NODE02
3. 3

   Test Failover

   Simulate a node failure by pulling the power on NODE01 while VMs are running. Within ~30 seconds, the cluster should restart the VMs on the surviving nodes automatically.

With AD in place, use GPO to enforce consistent configuration across all cluster nodes and VMs without manual per-machine work.

Useful GPO Configurations