Proxmox VE 9.2: Complete Guide and Upgrade

Proxmox Virtual Environment 9.2 is a major update to the open-source virtualization platform used in homelabs, private clouds and production clusters. Released on May 21, 2026, it combines a refreshed software stack with several features aimed specifically at larger clusters and more advanced network designs.

Proxmox VE 9.2 is based on Debian 13.5 “Trixie”, uses Linux kernel 7.0 as the stable default kernel and includes QEMU 11.0, LXC 7.0, ZFS 2.4 and Ceph Tentacle 20.2.

The version is much more than a package refresh. It introduces a Dynamic Load Balancer for HA-managed guests, expands the Software Defined Networking stack with native WireGuard and BGP fabrics, adds GUI management for custom CPU models and introduces an HA Arm/Disarm workflow for safer maintenance.

This guide explains the new features, installation, networking, storage, VM and LXC configuration, backup, security, cluster design and the upgrade path from earlier Proxmox VE releases.

What is Proxmox VE?

Proxmox Virtual Environment is an open-source virtualization management platform that brings virtual machines, Linux containers, storage, clustering, backup, networking and High Availability into one management plane.

KVM/QEMU virtual machines;
LXC system containers;
ZFS and local storage;
Ceph distributed storage;
Software Defined Networking;
cluster management and quorum;
High Availability;
integrated firewall;
backup and restore;
replication;
REST API and command-line tools.

The core functionality is available without a commercial subscription. A Proxmox subscription primarily provides access to the Enterprise repository and official support, while the no-subscription repository remains widely used for labs, testing and non-critical environments.

Proxmox VE 9.2 software stack

Component	Proxmox VE 9.2
Base distribution	Debian 13.5 “Trixie”
Default stable kernel	Linux 7.0
QEMU	11.0
LXC	7.0
ZFS	2.4
Ceph	Tentacle 20.2

Dynamic Load Balancer

The most visible cluster feature in Proxmox VE 9.2 is the new Dynamic Load Balancer. The Cluster Resource Scheduler can now use real-time node and guest resource utilization when making placement decisions for HA-managed workloads.

NODE01
CPU: 84%
RAM: 91%

NODE02
CPU: 27%
RAM: 34%

NODE03
CPU: 21%
RAM: 30%

In a static design, that imbalance may persist until an administrator manually migrates guests. With dynamic scheduling enabled, the cluster can automatically migrate HA resources to reduce node imbalance while respecting the HA rules defined by the administrator.

This matters most in clusters where dozens or hundreds of VMs compete for CPU and memory. It can reduce the amount of manual placement work required after maintenance, failover events or long periods of uneven workload growth.

WireGuard and BGP fabrics in SDN

Proxmox VE 9.2 significantly expands its Software Defined Networking capabilities. WireGuard and BGP are now available as fabric protocols, opening the door to encrypted multi-site connectivity and more advanced routed topologies.

Site A / Cluster A
+-------------------+
| PVE01      PVE02  |
+---------+---------+
          |
          | WireGuard fabric
          |
+---------+---------+
| PVE03      PVE04  |
+-------------------+
Site B / Cluster B

For larger environments, BGP can be used to integrate the Proxmox SDN layer with existing routing infrastructure. Route maps and prefix lists provide more granular control over which prefixes are accepted, filtered or redistributed.

These capabilities are especially useful for datacenters, cross-site virtualization clusters and environments where networking must scale beyond a single Linux bridge.

Custom CPU models from the GUI

Custom CPU profiles can now be managed directly from the Proxmox web interface. This is useful for clusters containing different CPU generations, where exposing the full host CPU to every VM may make live migration impossible.

Using CPU type: host generally offers the best access to host CPU features, but it also binds a VM more closely to the capabilities of that physical processor. In heterogeneous clusters, a carefully selected common CPU model is often a better operational choice.

The CPU flags selector in Proxmox VE 9.2 helps administrators verify which flags are supported across cluster nodes before creating a custom model.

HA Arm and Disarm for maintenance

Cluster maintenance can be risky when High Availability is active. A planned operation that temporarily affects a node or service can otherwise trigger automated HA actions at exactly the wrong moment.

Proxmox VE 9.2 adds an Arm/Disarm workflow for the HA Manager. Administrators can temporarily disarm HA across the cluster before maintenance and then arm it again afterward. HA resource state is preserved through the workflow.

Microsoft and Windows UEFI 2023 certificates

Proxmox VE 9.2 also improves support for modern Windows Secure Boot environments by supporting enrollment of Microsoft and Windows UEFI 2023 certificates through the GUI and API.

QEMU 11, LXC 7 and ZFS 2.4

The updated virtualization stack is important even for single-node deployments. QEMU 11 provides a newer virtual hardware base, LXC 7 updates the container stack and ZFS 2.4 continues to strengthen Proxmox as a platform for storage-heavy homelabs and private infrastructure.

VM or LXC?

Feature	Virtual Machine	LXC Container
Own kernel	Yes	No
Windows support	Yes	No
Isolation	Strong	Container-level
Resource overhead	Higher	Lower
Boot time	Slower	Very fast

For infrastructure services such as DNS, monitoring or lightweight Linux applications, LXC can be extremely efficient. For Windows, kernel-specific workloads, security-sensitive isolation or Docker platforms that should be kept independent from the hypervisor, a VM is usually the cleaner choice.

Installing Proxmox VE 9.2

For most installations, the official Proxmox VE ISO remains the simplest and most predictable method.

Download official ISO
        ↓
Verify checksum
        ↓
Create bootable USB / remote ISO mount
        ↓
Boot in UEFI mode
        ↓
Install Proxmox VE
        ↓
Configure static management IP
        ↓
Update packages
        ↓
Configure storage, network and backups

Hardware planning

64-bit CPU with hardware virtualization support;
IOMMU support if PCIe passthrough is required;
at least 16 GB RAM for a comfortable multi-VM homelab;
ECC memory for critical infrastructure when supported by the platform;
SSD or NVMe storage for active VM workloads;
separate backup storage;
reliable network interfaces;
UPS for systems that matter.

If ZFS is used, plan memory according to the real workload. Rules such as “1 GB RAM per TB” are not universal requirements. ARC size, VM memory, deduplication choices and the number of active datasets matter far more than a single generic ratio.

Basic network configuration

A basic installation normally uses a Linux bridge such as vmbr0. The physical NIC is attached to the bridge and VMs or containers connect to that bridge through virtual interfaces.

Physical NIC
    |
   vmbr0
    |
    +--- VM 100
    +--- VM 101
    +--- CT 200

Example:

auto lo
iface lo inet loopback

iface eno1 inet manual

auto vmbr0
iface vmbr0 inet static
    address 192.168.10.10/24
    gateway 192.168.10.1
    bridge-ports eno1
    bridge-stp off
    bridge-fd 0

Always adapt interface names, addresses and gateways to the actual environment. In production, separate management, storage, migration and tenant traffic where the network design justifies it.

Secure access to the web interface

The Proxmox web interface listens on TCP port 8006. Avoid exposing this management interface directly to the public Internet unless there is a very specific and well-controlled reason.

use a VPN;
place management on a dedicated VLAN;
use firewall allowlists;
enable multi-factor authentication;
use a bastion host where appropriate.

Update immediately after installation

apt update
apt full-upgrade

Then inspect the running versions:

pveversion -v
uname -r

Enterprise vs no-subscription repositories

A paid subscription is not required to unlock core Proxmox VE features. The important distinction is the update channel. The Enterprise repository is intended for subscription customers and receives packages that have gone through additional validation. The no-subscription repository moves faster and is commonly used in homelabs and testing environments.

Always confirm repository configuration before an upgrade:

apt update
apt policy

ZFS storage

ZFS remains one of the most attractive local storage options for Proxmox. It combines checksumming, snapshots, compression, datasets and replication in a storage layer that integrates closely with Proxmox VE.

zpool status
zfs list

Do not confuse RAID or a ZFS mirror with backup. Mirroring protects against certain disk failures, but deletion, ransomware, operator error and logical corruption can still affect the entire pool.

Ceph Tentacle 20.2

For distributed storage, Proxmox VE 9.2 ships with Ceph Tentacle 20.2 as the current stable option. Ceph is designed for a very different problem than a local ZFS mirror: it distributes data and availability across multiple storage nodes.

PVE01 -----+
           |
PVE02 -----+---- Ceph cluster
           |
PVE03 -----+

Ceph is powerful, but it is not automatically the best choice for every three-node homelab. Network latency, OSD count, drive class and failure domains all matter. A poorly designed Ceph cluster can perform worse than a much simpler local ZFS architecture.

Creating a Linux VM

use VirtIO networking;
use VirtIO SCSI for modern Linux guests;
install QEMU Guest Agent;
use UEFI when required by the guest or security design;
select the CPU model with migration requirements in mind.

On Debian or Ubuntu guests:

apt update
apt install qemu-guest-agent
systemctl enable --now qemu-guest-agent

Enable the QEMU Guest Agent option in the VM configuration as well.

LXC containers

Refresh the appliance template list with:

pveam update
pveam available

After downloading a template:

pveam list local

For most general-purpose containers, prefer an unprivileged LXC configuration unless the workload explicitly requires privileges that cannot be provided safely through mapped IDs and device permissions.

Docker on Proxmox

For important Docker workloads, a dedicated Linux VM is often cleaner than installing Docker directly on the Proxmox host. This keeps the hypervisor minimal, makes upgrades easier to reason about and simplifies backup and restore.

Backup and Proxmox Backup Server

Proxmox VE includes integrated guest backup through vzdump.

vzdump 100 \
  --storage backup \
  --mode snapshot

For larger environments, Proxmox Backup Server is the natural companion. It provides incremental backups, deduplication, retention management, verification, encryption and efficient restore workflows.

Back up the host configuration as well, especially:

/etc/pve/
/etc/network/interfaces
/etc/hosts
/etc/hostname
/etc/resolv.conf

Cluster design and quorum

A cluster can be created with:

pvecm create homelab

Additional nodes can join using the existing cluster address:

pvecm add 192.168.10.10

Understand quorum before building HA. With three voting nodes, a majority of two votes is required. Two-node clusters require special planning and may benefit from a QDevice depending on the architecture.

A Proxmox cluster is not automatically an HA design. HA also requires storage availability after node failure, valid quorum, sufficient spare compute capacity and correctly configured HA resources.

Upgrade from Proxmox VE 9.1 to 9.2

An upgrade inside the same major release is normally straightforward, but production systems should still be backed up and checked before package changes.

pveversion -v
apt update
apt full-upgrade

After the update, verify the active kernel and Proxmox packages and reboot when required:

uname -r
pveversion -v
systemctl --failed

Upgrade from Proxmox VE 8 to 9

The 8-to-9 upgrade is a major operating-system transition from Debian 12 Bookworm to Debian 13 Trixie and should be treated accordingly. Do not perform a blind repository search-and-replace followed by a dist-upgrade.

Proxmox provides the pve8to9 checker:

pve8to9 --full

Fully update the current Proxmox VE 8 host.
Verify cluster and storage health.
Back up all important VMs and containers.
Test at least one restore.
Back up /etc/pve and network configuration.
Check third-party repositories and packages.
Run pve8to9 --full.
Resolve every relevant failure before the upgrade.
Read the current official Proxmox upgrade documentation.
Ensure physical or out-of-band console access is available.

Post-upgrade checks

pveversion -v
uname -r
systemctl --failed
pvecm status
pvesm status
qm list
pct list

If ZFS is used:

zpool status
zfs list

If Ceph is used:

ceph -s

ARM64 availability

Proxmox now also publishes an ARM64 installer for Proxmox VE 9.2. ARM deployments should still be planned around the specific hardware support requirements of the platform, and migration remains architecture-dependent. Check the current Proxmox documentation and download notes before treating an ARM system as a production target.

Security checklist

Do not expose port 8006 unnecessarily.
Enable MFA for administrator accounts.
Use API tokens with minimum required permissions.
Keep the host updated.
Separate management traffic where possible.
Protect backups from the same failure domain as production.
Monitor failed services, storage health and cluster quorum.
Use firewall rules at Datacenter, node and guest level carefully.
Maintain out-of-band access for critical systems.

Useful diagnostic commands

pveversion -v
pvecm status
pvesm status
qm list
pct list
systemctl --failed
journalctl -p warning -b
ip -br addr
ip route
ss -lntup
zpool status

Proxmox VE 9.2 for a homelab

For homelabs, Proxmox VE remains one of the strongest open-source options because it lets a single physical server host multiple isolated services while still providing snapshots, backups, virtual networking and a clean management interface.

Proxmox host
   |
   +-- Firewall VM
   +-- Docker VM
   +-- DNS LXC
   +-- Monitoring LXC
   +-- Home automation VM
   +-- Linux / Windows lab VMs

Proxmox VE 9.2 for production

Production environments should treat the hypervisor as critical infrastructure. Redundant networking, tested backups, restore drills, monitoring, change control, documentation and out-of-band access matter at least as much as the hypervisor version itself.

Final checklist

Install from the current official Proxmox VE 9.2 image.
Apply all updates after installation.
Verify repository configuration.
Use a static management address.
Verify hostname and DNS.
Configure storage deliberately.
Check ZFS or Ceph health.
Configure external backups.
Test restore procedures.
Enable MFA.
Restrict management access.
Use QEMU Guest Agent where appropriate.
Plan CPU models for live migration.
Understand quorum before enabling HA.
Monitor hosts and storage.
Back up /etc/pve.
Document recovery procedures.

Is Proxmox VE 9.2 worth the upgrade?

For existing Proxmox VE 9.x deployments, 9.2 is a natural and important update. Cluster users benefit the most from the Dynamic Load Balancer, HA maintenance workflow and expanded SDN stack, while standalone hosts benefit from the refreshed Debian, kernel, QEMU, LXC and ZFS foundation.

For Proxmox VE 8 installations, the move to 9.x should be planned as a major operating-system upgrade rather than a routine package update. Backups, restore testing and the official pre-upgrade checker are essential.

Conclusion

Proxmox VE 9.2 is one of the most significant updates in the 9.x series. Dynamic workload balancing, native WireGuard and BGP fabrics, custom CPU model management and the HA Arm/Disarm workflow all target real operational problems found in larger virtualization environments.

At the same time, the updated Debian 13.5, Linux kernel 7.0, QEMU 11, LXC 7 and ZFS 2.4 stack makes it a compelling platform for both advanced homelabs and production virtualization.

Proxmox VE 9.2: Complete Guide, New Features, Installation and Upgrade