A step-by-step guide to building a two-node Raspberry Pi cluster running Proxmox VE (PXVirt), with a third machine acting as a QDevice tiebreaker for quorum.

Hardware Used

- 2x Raspberry Pi 4B (one 4GB, one 8GB)

- 1x x86 machine running Proxmox VE (used as QDevice)

- Network switch

Software

- PXVirt 9.0.10-2 (Proxmox for ARM/Pi)

- Proxmox VE 9.x (x86 tiebreaker node)

Install PXVirt on Both Raspberry Pis

Download and flash PXVirt to SD cards for each Pi (or a USB drive which I used). Set static IPs or ensure DHCP reservations are in place. In this guide:

- `warren-pi` = 192.168.1.10

- `chunguz-deluxe` = 192.168.1.199

- `michelle` (x86 QDevice) = 192.168.1.96

Fix the Hosts File

On each Pi, ensure the hostname resolves correctly:

1nano /etc/hosts
Copy

Make sure it contains:

```

192.168.1.10    warren-pi

192.168.1.199   chunguz-deluxe

```

Create the Cluster on Node 1

SSH into the first pi (in my case warren-pi) and create the cluster:

1pvecm create warren-pi
Copy

Verify it was created:

1pvecm status
Copy

Join Node 2 to the Cluster

SSH into the second pi (in my case chunguz-deluxe) and join the cluster:

1pvecm add 192.168.1.10
Copy

Verify both nodes are visible:

1pvecm status
Copy

You should see both nodes listed with votes.

Set Up QDevice on the x86 Machine

A two-node cluster has no quorum majority. A QDevice provides a third vote without being a full cluster member. This is useful especially if you don't have a third arm device, so I can use my x86 proxmox machine to be the QDevice.

On the x86 machine:

Install the QDevice daemon:

1apt install -y corosync-qnetd
2systemctl enable corosync-qnetd
3systemctl start corosync-qnetd
Copy

On the first pi (warren-pi for my example): Add the IP address

1pvecm qdevice setup 192.168.1.96
Copy

Verify the QDevice is active:

1pvecm status
Copy

You should see output like:

1Nodeid      Votes    Qdevice Name
2
30x00000001      1    A,V,NMW 192.168.1.10 (local)
4
50x00000002      1    A,V,NMW 192.168.1.199
6
70x00000000      1            Qdevice
8
9Expected votes: 3, Quorate: Yes ✅
Copy

Configure High Availability for VMs

To make a VM automatically restart on another node if a node fails, add it to the HA manager:

1ha-manager add vm:201 --state started --max_restart 3
Copy

Check HA status:

1ha-manager status
Copy