How to Set Up High Availability in a Proxmox Cluster

A single Proxmox host means a single point of failure — if that machine’s motherboard, power supply, or storage dies, every VM running on it goes down with it. Setting up Proxmox High Availability across multiple nodes means your VMs automatically migrate to healthy hardware when a node fails, keeping critical services running with minimal interruption.

What Is High Availability in Proxmox?

Proxmox’s HA feature continuously monitors the health of cluster nodes and configured VMs. If a node becomes unreachable or a monitored VM crashes, Proxmox automatically restarts that VM on another available node in the cluster, without requiring manual intervention — turning what would otherwise be a service outage into a brief automatic recovery.

Requirements Before Setting Up HA

• At least 3 nodes – Proxmox HA requires a minimum of three nodes to maintain proper quorum (more on this below); a two-node cluster cannot safely determine which node should take over during a failure
• Shared or replicated storage – VMs need their disks accessible from multiple nodes, either through shared storage (NFS, Ceph, iSCSI) or ZFS replication between nodes
• A stable, low-latency network between all cluster nodes, since HA relies on constant communication to determine node health

Understanding Quorum

Quorum is the mechanism Proxmox uses to prevent “split-brain” scenarios, where a network partition could otherwise cause multiple nodes to simultaneously believe they should take over the same VM. A cluster needs more than half its nodes actively communicating to maintain quorum; if quorum is lost, HA-protected VMs won’t be automatically migrated, avoiding potential data corruption from two nodes running the same VM simultaneously.

Setting Up a Basic Cluster

If you haven’t already formed a cluster, create one from your first node:

pvecm create my-cluster

Join additional nodes from each of their own consoles:

pvecm add <first-node-ip>

Verify all nodes have joined successfully:

pvecm status

Configuring Shared Storage

For HA to work, VM disks need to be reachable regardless of which node the VM ends up running on. Common approaches include:

• Ceph – a distributed storage system built into Proxmox, providing redundant storage across all cluster nodes
• ZFS replication – periodically replicates VM disks between specific node pairs, offering a lighter-weight alternative to full Ceph deployment for smaller clusters
• NFS/iSCSI – an external storage server providing shared access, simpler to set up but introducing that storage server as its own potential single point of failure

Adding a VM to HA Management

1. In the Proxmox web UI, go to Datacenter → HA.
2. Click Add under the Resources section.
3. Select the VM or container you want under HA protection.
4. Choose a Group (optional, for controlling which nodes a VM is allowed to run on) and Max Restart / Max Relocate settings.

Testing Failover

Before relying on HA in a real scenario, deliberately test it: with a non-critical test VM under HA protection, power off the node it’s currently running on (or disconnect its network) and confirm Proxmox correctly detects the failure and restarts the VM on another available node within the expected timeframe.

Understanding Failover Limitations

HA protects against node failure, but it isn’t instantaneous — there’s typically a delay of around a minute or two before Proxmox confirms a node is genuinely down (rather than experiencing a brief network hiccup) before initiating failover, meaning brief service interruption is expected, not eliminated entirely.

Is HA Worth It for a Home Lab?

HA adds real complexity — shared storage, network requirements, and a minimum three-node cluster — that’s genuinely worth it primarily for services you actually depend on running continuously (DNS, home automation, critical shared storage). For test VMs, disposable environments, or services where occasional downtime is a minor inconvenience rather than a real problem, the added complexity of HA may not be worth the overhead for most homelab setups.

Final Thoughts

Configuring Proxmox High Availability brings genuine enterprise-grade resilience to a homelab, automatically recovering from node failures that would otherwise take down critical services. While the setup requires more nodes and shared storage infrastructure than a simple single-host setup, for homelab users running services their household actually depends on, HA transforms a single point of failure into a genuinely resilient system.