Skip to Content

Using a Raspberry Pi as a Corosync QDevice for Proxmox HA Clusters

Learn how to turn a Raspberry Pi into a Corosync QDevice to meet the quorum requirement of a Proxmox HA cluster, with step‑by‑step commands and best‑practice tips.
9 February 2026 by
TechStora Editorial Board

Why a Raspberry Pi Can Save Your HA Cluster

Even though a Pi cannot run Proxmox VMs efficiently, it can act as an external arbitrator for the cluster’s quorum. This cheap, low‑power device lets you meet the three‑node requirement without buying additional x86 hardware.

Understanding the Quorum Requirement

Proxmox (and any Corosync‑based cluster) needs an odd number of voting members—typically three or more—to avoid split‑brain scenarios. If one node fails, the remaining nodes must still have a majority of votes, otherwise the web UI becomes inaccessible.

Setting Up the QDevice on the Raspberry Pi

Follow these commands on the Pi and each Proxmox node. All steps assume you have sudo access and the nodes can reach each other over the network.

  • On the Raspberry Pi, install the QNet daemon:
    sudo apt update && sudo apt install corosync-qnetd
  • On every Proxmox node, install the QDevice package:
    apt install corosync-qdevice
  • If a stale node still exists in the cluster, remove it:
    pvecm delnode <old-node-name>
  • Configure the QDevice, replacing <PI_IP> with the Pi’s IPv4 address:
    pvecm qdevice setup <PI_IP>
  • Accept the fingerprint prompt by typing yes.
  • Verify the setup on any node:
    pvecm status – the Pi should appear as “QDevice”.

Testing the Configuration

Power off one of the primary Proxmox nodes. The cluster should remain online, and the Pi will cast the tie‑breaker vote. Run pvecm status again to see the quorum count stay above the required threshold.

Limitations and When to Consider Alternatives

The QDevice works well for clusters that primarily use ZFS replication. However, if your nodes differ greatly in hardware or you plan to run a Ceph storage cluster, you’ll need a full third Proxmox node because Ceph requires at least three OSDs for data safety. Additionally, make sure HA rules are tuned so that a single node failure does not overload the remaining machines.