Automated Proxmox VE Node Install with Data Center Manager

Data Center Manager 1.1 can provision and join a fresh Proxmox VE node over the network in a single automated flow: you write one preseed file, point it at the target's NIC MAC address, and the node boots, installs, and registers itself without anyone at the keyboard. This guide walks through that flow end to end so you can stand up additional nodes in roughly ten minutes each instead of spending an afternoon on interactive installs. If you haven't read the piece on why you should try Data Center Manager 1.1 on your homelab, start there for the big-picture rationale; this is the how-to.

Key Takeaways

• One file drives it. A single preseed controls the OS install, hostname, network, and root password for every node in the run.
• MAC is the trigger. DCM boots the correct machine by matching the node's NIC MAC over PXE, so a wrong MAC is a silent failure.
• Ten minutes each. A fresh node goes from powered off to cluster-ready in roughly ten to twelve minutes.
• Homogeneous wins. Automation shines on identical hardware; custom partitioning still calls for an interactive install.

What Can Data Center Manager 1.1's Host Install Do?

Data Center Manager runs a control plane — a lightweight system container that sits above your nodes and manages them as a group. Version 1.1 added automated host installation, which means the control plane can drive a network boot of a brand-new node, feed it a preseed, and hand back a clean Proxmox VE install that is already registered in the data center inventory.

The flow replaces the interactive installer with a file. You define the install once in a preseed, attach that preseed to a node entry by its MAC address, and DCM handles the PXE boot, package download, partitioning, and registration for you. When it finishes, the node shows up in the control plane and you add it to your cluster with the usual commands.

This is most useful when you're adding several nodes that should look the same. It is less useful for a one-off box with unusual disks or a custom RAID layout, where the interactive installer still gives you more control. The control plane is what turns a pile of bare metal into a managed fleet, and automated installation is the part that scales that management out.

What Do You Need Before You Boot a Node?

Before anything boots, line these up:

• A Data Center Manager control plane installed and healthy on one of your existing nodes.
• A cluster already running, or at least a plan for the first node — you cannot join a node to a cluster that does not exist yet.
• The target node with network boot (PXE) enabled in the BIOS/UEFI, and connected to the same L2 segment as the control plane. PXE does not cross routers without a relay, so keep the control plane and the target on the same switch.
• The exact MAC address of the NIC that will boot. One wrong character and the node will not start the install.
• A reachable Proxmove mirror, plus the subnet, gateway, and nameserver details the node should use.

Get the networking right first, because a mistyped subnet will leave your fresh node unmanageable. That is exactly the kind of thing our VLAN vs VXLAN breakdown walks through. If you already use SDN zones to hand out addresses, fold that networking into the preseed so the node comes up on the right bridge from minute one.

Step 1: Install the Control Plane

The control plane runs as a system container (LXC), and the simplest path is the GUI. In the Proxmox VE web interface, open the Data Center Manager section and install the control plane there. Proxmox creates the container, pulls the template, and starts it — no manual container setup required.

Confirm the version and that the control plane is responding:

pveversion -v

You should see the Data Center Manager version line alongside your Proxmox VE version. If the GUI install has not happened yet, use the Data Center Manager install option rather than guessing at a command line — the container needs the right template and startup settings, and the GUI sets those correctly. A healthy control plane is the foundation everything else builds on, so do not skip this check.

Step 2: Write the Preseed File

The preseed is the heart of the whole flow. It is the standard Proxmove installer preseed, so anything the interactive installer accepts works here. Create a file like this:

# --- Proxmox VE preseed for Data Center Manager ---
# Mirror
d-i mirror/country string manual
d-i mirror/hostname string pve-mirror.proxmox.com
d-i mirror/path string proxmox
d-i mirror/scheme string https

# Packages
d-i base-installer/install-recommends string false

# Time
d-i time/zone string UTC
d-i clock-source string UTC

# Root password (generate with openssl passwd -6)
d-i passwd/root-password-crypted value $6$AbCdEfGh$IxYz0123456789abcdefghijklmnopqrstuv

# Hostname and network
d-i netcfg/hostname string pve-node2
d-i netcfg/get_hostname string pve-node2
d-i netcfg/get_ipaddress string 10.0.0.12
d-i netcfg/get_netmask string 255.255.255.0
d-i netcfg/get_gateway string 10.0.0.1
d-i netcfg/nameservers string 10.0.0.2

# Partitioning: wipe and install to the first disk
d-i partman/choose_partition select finish
d-i partman/confirm boolean true
d-i partman/confirm_write_new_label boolean true

# Boot loader
d-i grub-installer/only_debian boolean true

The one field that trips people up is the root password. It has to be a hash, not plain text, or the install fails. Generate one on any existing node:

openssl passwd -6 'S3cureP@ss'

Paste the $6$ result into the passwd/root-password-crypted line. If you are templating many nodes, give each node a different root password and keep the hashes in the per-node preseed.

A couple of things about the file itself: it stays small — a lean preseed is a few hundred bytes to about a kilobyte — so you can keep it under version control and diff it like code. And because the network block lives here, this is where your VLAN or SDN decisions land; get the IP, netmask, gateway, and nameserver correct once and every node inherits them.

Step 3: Register the Node and Trigger the Install

Back in the Data Center Manager GUI, open the Nodes view and create a new node. Fill in the name, then enter the MAC address of the NIC that will PXE boot — this is the address DCM matches against when it starts the install. Attach the preseed file you just wrote, then start the job.

The control plane hands the preseed to the node over the network and watches the boot. You can watch progress in the same Nodes view; the row flips through states as the download, partitioning, and package install run. On a typical rack server this whole stretch takes roughly ten to twelve minutes, mostly the package download.

If the node never leaves the "booting" state, the first thing to check is the MAC address — a single transposed character is the most common reason a provisioning job stalls. The second check is connectivity: the target and the control plane must reach each other on the same L2 segment for PXE to work.

Step 4: Let the Node Boot and Join

When the install finishes, the node is registered in the control plane and booted into a clean Proxmox VE. It is not yet part of your cluster, though. Grab its IP and SSH in as root, then add it to the cluster:

pvecm add 10.0.0.10 --fingerprint 9F:2A:7C:1E:4B:8D:3A:6F:5C:0E:7B:9D:2A:4F:8E:1C:6B:3D:7A:0F

The fingerprint is the one shown for the node you are joining in the Proxmox VE GUI, and passing it with --fingerprint skips the interactive trust prompt. If you are scripting several nodes, you can put this same command into a post-install step so the join happens without touching the console.

Before the node is genuinely useful, it needs a storage layout, because the freshly installed box has nothing on it yet. Decide between local storage and cluster storage up front — the ZFS pools and datasets guide is the right place to start for local, and the ZFS vs Ceph comparison helps you pick cluster storage before the node is even added.

How to Verify the Installed Node

Once the join completes, confirm both the cluster membership and the control-plane view:

pveversion -v
pvecm status
pvecm nodes

pveversion -v confirms the installed Proxmox VE version, pvecm status shows the quorum and node list, and pvecm nodes lists every node the cluster can see. In the Data Center Manager GUI the node should read as healthy and online. If pvecm nodes shows your new node as offline, check the network settings in the preseed — a wrong gateway or nameserver is the usual culprit.

What's the Catch? Gotchas and Tradeoffs

Automated installation is not magic, and a few things bite.

The biggest gotcha is the MAC address. DCM matches the preseed to hardware by MAC, and there is no fallback — a typo just means the node does nothing. The second is the network. PXE stays on one L2 segment, so your control plane and every target node have to be reachable without a relay, or the boot never starts. The third is the password: it has to be a real hash, and a plain-text password silently fails the install.

The honest tradeoff is flexibility. Automation wins when your nodes are identical and you want them to look the same. It gets awkward when hardware varies or you need a custom partition or RAID layout, because the preseed has to express all of that in advance. Compare the two head to head:

Aspect	Automated (DCM preseed)	Interactive installer
Hands-on time per node	~10 min, unattended	30–60 min at the keyboard
Best for	Uniform node fleets	One-off or mixed hardware
Custom partitioning / RAID	Limited, preseed-driven	Full manual control
Network requirement	PXE on one L2 segment	Any bootable media
Repeatability	High — one file drives all	Manual, every time

For a homelab or a small cluster of identical boxes, the automated path is worth the upfront preseed work. For a mixed fleet, keep the interactive installer in your back pocket.

Conclusion

Data Center Manager 1.1's host install turns node provisioning into a file-driven flow: write a preseed, match it to a node by MAC, and watch a fresh Proxmox VE box boot, install, and register in about ten minutes. The payoff is real for homogeneous fleets, and the cost is just getting the network and preseed right up front. Your next step is to provision one spare node end to end before you batch the rest — catch any MAC or subnet mistakes on a single box, then repeat.