masalladelcloud
Yesterday, Proxmox VE 9.1 was released, and it’s not one of those “minor” releases where there’s hardly anything new. This is the first point release after the major leap to version 9.0, and frankly, if you’re in the infrastructure world, there are things here that will interest you quite a bit.
Analysis Objective: Review the critical improvements introduced in Proxmox VE 9.1, including support for OCI containers, TPM snapshots, and network upgrades. We will evaluate the real-world impact of these features in homelab and production environments.
Key New Features
| Feature | Production Impact |
|---|---|
| OCI Containers | Native deployment of container images without Docker overhead |
| TPM Snapshots | Full backups of Windows 11 VMs without breaking Secure Boot |
| Integrated SDN | Greater visual control over complex virtual network topologies |
I’ve been using Proxmox for a long time, and this update feels like someone finally understands what those of us who manage machines at home or on a small scale need. Let’s see what it brings.
Game Changer: OCI Containers Directly in LXC
Remember when you had to jump through hoops to create container templates? Well, that has changed.
Now in 9.1, you can take standard OCI images (the ones you use on Docker Hub, official Linux repo ones, whatever) and deploy them directly as LXC templates. No need to convert anything, no weird scripts. You simply pull from a registry or upload the image manually, and that’s it.
Interstingly, Proxmox is smart enough to know what type of container you need: whether it’s a full system container or a lighter one, optimized for microservices. The latter are very cool if what you want is to deploy something specific without all the overhead of a full system.
Why does this matter to you? Because it means you can reuse images you already have, that you already know, without reinventing the wheel every time. It’s simply more efficient.
Windows VMs with TPM Can Finally Take Snapshots
This is something many have been asking for for a long time: being able to take snapshots of virtual machines with TPM (that security chip Windows 11 and new versions use). Before, you couldn’t, and you had to do weird balancing acts.
In 9.1, the TPM state is saved directly in the qcow2 file. This means you can now:
- Take full snapshots of your Windows 11/Secure Boot VMs without issues.
- Move those snapshots across network storage (NFS, CIFS) without anything breaking.
- Properly back up machines that depend on TPM.
If you have a home lab with several Windows machines, this is a considerable relief. You no longer have to think about workarounds.
Finer Control of Nested Virtualization
Those who play with nested virtualization (running hypervisors inside VMs) will appreciate this. Now you have a new vCPU flag that allows you to control exactly which virtualization extensions you expose to the virtual machine, instead of just copying the host’s entire CPU.
What is it for? For things like running Windows Virtualization-based Security inside a VM, or if you need to run a mini-hypervisor for testing. It’s finer, more controllable, and generally more efficient because you’re not exposing everything unnecessarily.
Networking (SDN) is Now Much More Transparent
Proxmox has been improving its software-defined networking stack for years, but in 9.1 they finally brought real visibility to the web interface.
Now you can see in the GUI:
- Which machines are connected to each bridge or VNet.
- In EVPN configurations, which IPs and MACs the network has “learned”.
- Routes, neighbors, interfaces (all integrated into the resource tree).
If you’re one of those who used to have to access via terminal to do network debugging, this is a change for the better. Especially if you have complicated topologies.
Batch Actions at the Datacenter Level
Small thing, but very useful: you can now select a group of virtual machines and perform batch operations on all of them. Massive restart, joint shutdown, batch migrations.
When you have 20-30 VMs in a home lab and need to perform maintenance, this saves you hours of clicking.
Under the Hood: Updated Versions
Proxmox 9.1 runs on Debian 13.2 “Trixie” with kernel 6.17. It’s not a trivial change. It features:
- QEMU 10.1.2 — better performance in VMs, better live migration.
- ZFS 2.3.4 — more stable snapshots and clones, improvements in scrub/resilver.
- LXC 6.0.5 — better container management.
- Ceph Squid 19.2.3 — for those with distributed storage.
If you’re coming from 8.4, the jump is quite large. And it’s a good jump.
A Note on Kernel 6.17
Here’s some advice: although 9.1 is generally stable, some people are reporting issues with kernel 6.17 on specific hardware (NVIDIA vGPU, some drivers, old machines). If you have “special” or custom hardware, maybe it’s worth testing in a lab first or keeping the kernel pinned to a previous one.
Should I Update?
Short answer: yes, probably.
If you’re on 8.4 and your infrastructure is stable, you’re in no hurry. But if you want the new features —especially OCI containers in LXC, TPM snapshots, or simply to be up to date— 9.1 is the point release where it’s definitely stable and ready for production.
To update from 8.4, you have to do it step by step: first to 9.0, then to 9.1. There’s no direct 8.4→9.1 path, and that’s for process security reasons (the switch to Debian 13 is not trivial).
The important thing: make a backup. Always. Although Proxmox is quite solid with its upgrades, having a Plan B never hurts.
Troubleshooting and Technical Considerations
Plugin Incompatibility After Update Review any third-party scripts or plugins (like dark theme scripts) before updating to 9.1, as the underlying Sencha UI changes may break them.
Performance in OCI Containers This is an emerging technology within Proxmox. For mission-critical workloads, evaluate keeping traditional LXCs or nested Docker until version 9.2 to ensure absolute stability.
Kernel 6.17 Regressions
If you have “special” hardware (NVIDIA vGPU, older drivers), some users report instability. If you fail to boot after updating, select the previous kernel from the GRUB/systemd-boot menu and pin it using proxmox-boot-tool.
So there you have it, now you know what’s special about 9.1 and why it’s worth paying attention to this update. If it’s been useful to you, share it with those administrator colleagues who are still on 8.4 without knowing what they’re missing. If this post has helped you, share it with other administrators and engineers who can benefit. And follow me for more real experiences from the homelab trenches.