Proxmox VE 9.2 Upgrade Guide — From 8.x with Dynamic Load Balancer

If your homelab runs Proxmox VE, May 21, 2026 marks a major release: Proxmox VE 9.2 ships with QEMU 11.0, LXC 7.0, ZFS 2.4, the Linux 7.0 kernel, and a brand-new Dynamic Load Balancer for automated cluster resource optimization.

This guide covers the full upgrade from Proxmox VE 8.4 to 9.2 — including the new features worth enabling after the reboot, known pitfalls like LXC containers failing to start on kernel 7.0, and the config changes you need to make for a clean upgrade on a homelab cluster.

Why Upgrade to Proxmox VE 9.2?

Proxmox VE 9.2 bundles several major updates under the hood:

The headline feature is the Dynamic Load Balancer, which automatically redistributes VMs and containers across cluster nodes based on CPU and memory utilization — similar to VMware DRS. Other notable additions include:

• SDN WireGuard fabrics — encrypted overlay networks between cluster nodes without dedicated VPN hardware.
• BGP/EVPN route maps and prefix lists — fine-grained control for advanced SDN setups.
• Custom CPU profiles in the web UI — create, edit, and remove CPU profiles per VM from Datacenter → Options.
• HA rules replacing HA groups — more flexible scheduling policies.

Prerequisites

Before touching the upgrade, verify these prerequisites:

1. Run the Latest Proxmox VE 8.4

1
2

pveversion
# Must report at least 8.4.1

If your version is older, update first:

1

apt update && apt dist-upgrade -y

2. Run the Pre-Upgrade Checklist

The pve8to9 script validates your system before the upgrade. It is included in the latest Proxmox VE 8.4 packages:

1

pve8to9 --full

Run it now. It will flag:

• LVM volumes with autoactivation enabled (common on LVM-thin storages)
• Third-party repositories that might conflict with Trixie
• Packages that will be removed during the dist-upgrade

Fix each warning before proceeding. Re-run pve8to9 --full after each fix.

3. Backup Everything

Do not skip this. Back up all VMs and containers to external or NFS/SMB storage via the GUI or:

1

vzdump <VMID> --storage backup-nfs --mode snapshot

Also back up critical host config:

1

tar czf /root/pve-etc-backup-$(date +%F).tar.gz /etc/pve /etc/network/interfaces /etc/hostname /etc/resolv.conf

4. Ensure Enough Disk Space

1
2

df -h /
# Need at least 5 GB free, ideally 10+ GB

The dist-upgrade downloads hundreds of packages. The apt cache alone can reach 1-2 GB during the process.

5. Use tmux or screen

If connecting via SSH (not IPMI/iKVM), wrap the session:

1

tmux new -s pve-upgrade

This protects against connection drops mid-upgrade.

Step-by-Step: In-Place Upgrade to Proxmox VE 9.2

Step 1: Update Repositories to Debian Trixie

Switch Debian base and Proxmox repositories from Bookworm to Trixie:

1
2
3
4
5

# Update Debian base repos
sed -i 's/bookworm/trixie/g' /etc/apt/sources.list

# Update Proxmox enterprise or no-subscription repo
sed -i 's/bookworm/trixie/g' /etc/apt/sources.list.d/pve-enterprise.list 2>/dev/null || true

Step 2: Add Proxmox VE 9 Repositories (deb822 Format)

Replace the old .list files with the new deb822 .sources format. For the no-subscription repository (most homelabs):

1
2
3
4
5
6
7

cat > /etc/apt/sources.list.d/proxmox.sources << 'EOF'
Types: deb
URIs: http://download.proxmox.com/debian/pve
Suites: trixie
Components: pve-no-subscription
Signed-By: /usr/share/keyrings/proxmox-archive-keyring.gpg
EOF

If you have an enterprise subscription:

1
2
3
4
5
6
7

cat > /etc/apt/sources.list.d/pve-enterprise.sources << 'EOF'
Types: deb
URIs: https://enterprise.proxmox.com/debian/pve
Suites: trixie
Components: pve-enterprise
Signed-By: /usr/share/keyrings/proxmox-archive-keyring.gpg
EOF

Remove the old .list files after verifying the new ones work:

1
2
3
4
5
6
7

apt update && apt policy
# Verify output shows trixie repositories

# Then remove old repo files
rm -f /etc/apt/sources.list.d/pve-enterprise.list
rm -f /etc/apt/sources.list.d/pve-install-repo.list
apt update

Step 3: Run the Dist-Upgrade

1

apt dist-upgrade

This upgrades the entire system from Debian 12 → 13 and Proxmox VE 8 → 9 in one pass. Expect 5-60 minutes depending on disk speed.

During the process you will be prompted about config file changes. Recommended choices:

Step 4: Reboot

1

reboot

After reboot, verify the kernel:

1
2
3
4

uname -r
# Should show: 7.0.x-pve
pveversion
# Should show: pve-manager/9.2.x

Step 5: Clean Up apt Repositories (Optional)

Modernize remaining .list files to deb822 format:

1

apt modernize-sources

This creates .sources files and renames old .list files to .list.bak. Verify everything works and delete the .bak files once confirmed.

Post-Upgrade: What to Fix First

LVM Autoactivation Migration

Upgrading from PVE 8, your existing LVM guest volumes may still have autoactivation enabled. This causes issues on shared LVM storages in clusters. Run the migration script:

1

/usr/share/pve-manager/migrations/pve-lvm-disable-autoactivation

Re-run pve8to9 --full to confirm the warning is gone.

LXC Containers Not Starting on Kernel 7.0

A common issue reported on kernel 7.0 is LXC containers failing to start with errors like:

lxc_map_ids: newuidmap failed to write mapping

This is caused by systemd-tmpfiles not creating /run/pve early enough. Fix:

1
2
3
4
5
6
7
8
9

systemctl status pve-guests.service
# Check for failures

# Ensure tmpfiles.d is configured correctly
systemd-tmpfiles --cat-config | grep pve

# If /run/pve is missing, create it and restart:
mkdir -p /run/pve
pve-container restart --all

For persistent issues, ensure pve-container and pve-guests services are enabled:

1
2

systemctl enable --now pve-guests.service
systemctl enable --now pve-container.service

Disable Audit Noise (Optional)

Debian Trixie enables kernel audit logging by default. To suppress excessive audit messages during normal operation:

1

systemctl mask --now systemd-journald-audit.socket

Configuring the Dynamic Load Balancer

The Dynamic Load Balancer is the standout feature in 9.2. It automatically rebalances VMs and containers across cluster nodes based on resource utilization — no more manually deciding where to place guests.

Prerequisites

• All cluster nodes must be on Proxmox VE 9.2.
• HA must be enabled on the cluster (Datacenter → HA).
• A valid quorum (at least 3 nodes or a QDevice).

Enable Dynamic Load Balancing

Navigate to Datacenter → Options and set:

• HA Scheduling: Dynamic Load
• Re-balance method: Default (bruteforce) or TOPSIS
• Automatically re-balance HA resources: ✓ enabled

From the CLI:

1
2
3
4

pvesh set /cluster/ha/options \
  --scheduler dynamic \
  --rebalance-method bruteforce \
  --rebalance-automatic 1

The load balancer periodically evaluates node utilization and moves low-priority VMs from overloaded nodes to underutilized ones. Migration thresholds are configurable:

1
2
3
4
5

pvesh set /cluster/ha/options \
  --cpu-threshold-overloaded 80 \
  --memory-threshold-overloaded 85 \
  --cpu-threshold-underloaded 30 \
  --memory-threshold-underloaded 40

HA Rules vs. HA Groups

Proxmox VE 9 deprecates HA groups in favor of HA rules. If you upgrade from PVE 8 with existing HA groups, they are automatically migrated to HA rules once all cluster nodes are on PVE 9.

Verify the migration:

1

journalctl -eu pve-ha-crm

Create new rules from the GUI or CLI:

1
2
3
4
5
6

pvesh create /cluster/ha/rules \
  --id my-rule \
  --type resource \
  --match "tags:critical" \
  --action spread \
  --priority 100

Other New Features Worth Enabling

SDN WireGuard Fabrics

Proxmox VE 9.2 adds WireGuard as a native SDN fabric protocol. This creates encrypted overlay networks between cluster nodes without configuring WireGuard manually on each host.

Datacenter → SDN → Zones → Add → Fabric

Select WireGuard as the fabric type, assign a subnet, and select which nodes participate. The zone is automatically provisioned and key exchange is handled by the cluster configuration.

Custom CPU Profiles

Navigate to Datacenter → Options → CPU Profiles to create custom CPU type profiles. Useful for:

• Exposing specific CPU features to VMs (e.g., AVX-512 for AI workloads)
• Matching CPU models across heterogeneous cluster hardware
• Hiding features for cross-cluster migration compatibility

BGP/EVPN Route Maps

If you run BGP/EVPN in your homelab (for example, integrating Proxmox SDN with RouterOS or FRR), the new route maps and prefix lists give you fine-grained control over route advertisement. Configure under Datacenter → SDN → IPAM / Route Maps.

Known Issues and Workarounds

Veeam Backup Compatibility

PVE 9.2 uses QEMU machine version 10.0+ by default. Veeam has not adapted to the new internal disk attachment changes. If you use Veeam for backups, pin affected VMs to machine version 9.2+pve1:

1

qm set <VMID> --machine pc-q35-9.2+pve1

/tmp is Now tmpfs

Debian Trixie mounts /tmp as tmpfs using up to 50% of system memory by default. If your workflows depend on large temp files, adjust the mount:

1

mount -o remount,size=10G /tmp

Or revert to disk-backed /tmp:

1

systemctl mask tmp.mount

Ceph Updates

If you run hyper-converged Ceph, the default is now Ceph Tentacle 20.2.1. Ceph Squid 19.2.3 remains available as an alternative. Check compatibility before upgrading:

1

ceph --version

Verifying a Clean Upgrade

Run the checklist once more after reboot:

1

pve8to9 --full

Expected clean output — no warnings or errors. Then:

1
2
3
4
5
6

pveversion -v | head -10
# pve-manager: 9.2.x
# proxmox-ve: 9.2.x
# proxmox-kernel-7.0: 7.0.x
# qemu-server: 11.x
# lxc-pve: 7.x

Check all VMs start properly:

1
2

qm list
pct list

Test HA rebalancing by creating a test VM with HA enabled, loading it with stress, and verifying the dynamic load balancer moves it:

1
2
3
4
5
6

# Install stress in a test VM
apt install stress
stress --cpu 4 --timeout 120

# Monitor load balancer activity
journalctl -fu pve-ha-crm

Summary

Proxmox VE 9.2 is a significant release for homelab users. The Dynamic Load Balancer eliminates the manual guesswork of VM placement, SDN WireGuard fabrics simplify encrypted cluster networking, and the underlying upgrade to Debian Trixie with Linux 7.0 brings modern kernel features.

Key takeaways:

1. Run pve8to9 --full before and after the upgrade.
2. Use tmux — this is a long-running dist-upgrade.
3. Fix LVM autoactivation post-upgrade to avoid storage issues in clusters.
4. Enable the Dynamic Load Balancer after all nodes are upgraded.
5. Pin Veeam-backed VMs to machine version 9.2+pve1 if needed.
6. Test LXC containers after the kernel 7.0 upgrade and apply the tmpfiles fix if they fail to start.

The 9.2 release transforms Proxmox from a capable hypervisor into an actively load-managed virtualization platform — and for homelab operators managing multiple nodes, that alone is worth the upgrade.