Several similar servers in a rack can make a simple question surprisingly difficult: which physical machine corresponds to the Proxmox node I am looking at?

That was the starting point in my environment. After matching the nodes to their chassis, I wanted the useful identification details to remain visible directly in Node → Summary, rather than having to open a shell every time.

This guide describes the working modification we used: collect the manufacturer, model and serial number from the server firmware, store them in a marked section of the node’s Notes, and display them in the existing Summary information panel. A manually entered rack location completes the picture.

This is an unofficial local interface modification, not an official Proxmox plugin. It worked in our environment. The important distinction is between a practical, functioning customization and a promise of compatibility with every Proxmox release.

Downloads

The installer is the unchanged v2.1 file used for the original guide. Its dashboard labels are Slovak: Identifikacia servera means server identification, Seriove cislo means serial number, nezistene means unknown, and nezadane means not specified. The instructions below are in English; preserving the original installer also preserves its checksum and existing installation behaviour.

What the result looks like

An illustrative display from this version is:

Identifikacia servera - pve01
Server: HP ProLiant DL360p Gen8
Seriove cislo: nezistene
Rack: Rack A / U18

These are example values, not an inventory of a published production server.

The installation was used on HP ProLiant DL360 G7 and DL360p Gen8 hardware. One server did not provide a serial number through DMI. That does not prevent installation: version 2.1 explicitly records missing fields as unknown instead of inventing an identifier.

The original environment recorded Proxmox VE 9.2.11. Treat that as the environment reported during this case, not as a general compatibility guarantee. The installer checks for Proxmox VE major version 9 and specific interface structures; those checks do not prove compatibility with every build.

1. Check the identity of the physical node

Open the Shell of the physical Proxmox node, not a virtual machine console:

hostname
pveversion
dmidecode -t system

Look for Manufacturer, Product Name and Serial Number.

The dmidecode project documentation explains that the tool reports hardware information supplied through SMBIOS/DMI. This is firmware-reported inventory, not an independent verification of the hardware. Fields can be missing or inaccurate.

Do not silently replace a missing system serial number with an SKU, UUID or baseboard serial. If the exact identifier matters, compare it with the chassis label or the server’s management interface.

This modification helps identify a server. It does not diagnose why a node has frozen, and a missing identifier is not evidence of a failed host.

2. Prepare access independent of the web interface

Before modifying the interface, verify that SSH or a physical console works. A broken Summary panel must not leave the web GUI as your only route to recovery.

From Windows Terminal or PowerShell, for example:

ssh root@IP_PROXMOXU

Replace IP_PROXMOXU with the server address and verify the SSH host-key fingerprint. Enter credentials directly into your SSH client, never into the article or the script.

You need root access on the physical node, Python 3, dmidecode, pvesh and a compatible Proxmox VE 9 installation. The script uses Python’s standard library; no pip packages are required.

3. Transfer and inspect the installer

Download the installer above and transfer it using SFTP, for example with WinSCP, to:

/root/pve-rack-id-v2.1.py

Read the source before running it as root. Do not pipe an unreviewed download directly into a shell.

Check the transferred file:

sha256sum /root/pve-rack-id-v2.1.py

The SHA-256 of the downloadable original file is:

e28832c4d972bcf9ebada1f168ff7339a540e3a1914e8f6dc048fbab41831877

This checksum compares file integrity with this release. It is not a digital signature and does not authenticate a separately modified copy.

4. Install the Summary modification

On the selected physical node, run:

python3 /root/pve-rack-id-v2.1.py install

The installer:

  1. Checks the expected template and Summary component structures.
  2. Collects the local node name and DMI identification fields.
  3. Updates its marked JSON section in Notes while preserving unrelated text.
  4. Creates backups under /root/pve-rack-id-backups/ before its changes.
  5. Writes a separate JavaScript file, /usr/share/pve-manager/js/pve-rack-id-v2.js.
  6. Adds a script reference to /usr/share/pve-manager/index.html.tpl.
  7. Reads back Notes and the written interface files to verify their contents.

The node description is a supported configuration field, documented in Proxmox node management and the pvenode command reference. Using that field for the inventory does not make the custom Summary display an official integration.

The large pvemanagerlib.js bundle is inspected but not rewritten. The JavaScript filename remains v2 because the v2.1 correction concerns incomplete inventory collection rather than a new interface format.

Installation is not a single transaction: Notes are updated before the interface files. If the later interface step fails, the collected inventory can remain in Notes. Stop and inspect the reported error rather than assuming that every earlier step was rolled back.

After successful installation, use Ctrl+F5, open the node → Summary, and check the identity display alongside the original statistics, menus and graphs. The script does not restart services, virtual machines or the host, and it does not change storage or networking configuration.

5. Add the rack location

Replace the example with the actual location:

python3 /root/pve-rack-id-v2.1.py collect --rack "Rack A / U18"

Refresh Summary to see the updated record.

To refresh firmware-derived inventory while keeping the existing rack value:

python3 /root/pve-rack-id-v2.1.py collect

This is a snapshot from the last collection, not a live hardware inventory feed. Run collect again after replacing hardware or updating its recorded location.

6. Multiple nodes: collection and interface installation are different

Collect identity locally on every physical node you want to identify. The display requests /nodes/{selected-node}/config and rejects an identity record whose node name does not match the selected node.

Install the interface modification on the nodes whose web addresses you use. Changing local web files on one host does not install them on every host in the cluster.

The simplest approach is to transfer the installer and run install on each intended node. The tool deliberately does not perform an automatic cluster-wide rollout.

Existing Notes and inventory remain subject to Proxmox API permissions. This modification does not add a separate authentication mechanism or bypass access checks.

7. Remove the modification

While using the compatible Proxmox VE 9 environment, run through SSH or a physical console:

python3 /root/pve-rack-id-v2.1.py remove

Refresh the browser with Ctrl+F5. The command removes its template hook and JavaScript file. The inventory in Notes is deliberately retained.

It does not blindly restore an old complete template: after an update, that template might belong to an incompatible version.

Recovery if the normal removal command cannot run

The installer’s major-version check also applies to remove. After a future major upgrade, or if the installer is unavailable, use SSH or a physical console to inspect the current template:

cp -a /usr/share/pve-manager/index.html.tpl /root/index.html.tpl.before-rack-id-removal
nano /usr/share/pve-manager/index.html.tpl

Remove only the script line referring to:

/pve2/js/pve-rack-id-v2.js?v=2

Save the file and hard-refresh the browser. Check that the original interface works. If an update already removed that line, do not delete another script reference or restore an old template just to match this guide.

Once the custom hook is absent, the unreferenced pve-rack-id-v2.js file is no longer loaded by that template. Inspect it before any manual cleanup. Never delete a different or independently modified extension without understanding it.

If the installer reports a different existing JavaScript file, stop rather than forcing an overwrite. An older, broken v1 installation should be removed using its original removal procedure before installing this version.

What we corrected during development

The first attempt called getHeight() during initComponent, before the component’s underlying element existed. It produced:

Cannot read properties of undefined (reading 'getHeight')

The working correction sets a numeric height configuration before parent initialization rather than measuring a nonexistent element.

During the original work, we reproduced the error in an isolated browser harness using the server’s actual Ext JS and Proxmox static libraries. API responses in that harness were simulated. The corrected panel rendered there, and the working display was subsequently confirmed in the real environment. These are different kinds of evidence; neither is a compatibility test for every Proxmox version.

The second correction removed an unnecessary installation failure when the firmware supplied a blank serial number. Version 2.1 records nezistene instead. A command failure, such as dmidecode exiting unsuccessfully or an API write being denied, is still an error rather than a successful partial inventory.

Before publishing this guide, the five Python regression tests were rerun successfully. They cover Notes preservation, an idempotent and reversible template hook, file installation/removal, digest-aware collection with read-back, and a whitespace-only serial number. These use temporary files and simulated commands; they do not constitute a new live-host or browser test.

Troubleshooting and maintenance

  • After a pve-manager update: recheck the hook, component compatibility and actual rendering. Package updates can overwrite or invalidate a local customization.
  • Blank or broken panel: capture the first browser-console error, remove the custom hook through independent access and verify that the original interface returns. Do not reboot guests or change cluster settings for a display problem.
  • Missing data: run collect on the selected physical node. Confirm that the displayed node is the intended one.
  • API access denied: check the account’s relevant Proxmox permissions. Do not broaden permissions merely to hide an error message.
  • DMI hardware-failure warning: investigate it separately through the hardware logs. The inventory modification does not repair hardware faults.
  • Screenshots: redact real serial numbers, internal addresses and internal system names before publication.

If Windows SSH rejects a local .ssh/config file because of ownership or permissions, a one-off connection can skip configuration-file loading:

ssh -F none root@IP_PROXMOXU

As documented in the OpenSSH manual, -F none suppresses configuration-file loading. It does not repair file permissions or disable host-key verification. Do not use it blindly if the connection depends on settings in those files.

The operational lesson

This is a small modification, but it solves a real usability problem: when I open a node, I can immediately see which physical server I am managing and where it belongs in the rack.

The modification is not a monitoring system, a health diagnosis or a substitute for chassis labels. It is a practical bridge between the logical node name and the physical machine, with explicit limits and a removal path.

For wider platform context, see the Proxmox VE Research profile. Related operational notes cover Windows migration to Proxmox with Veeam and Docker networking after a Hyper-V to Proxmox migration.

References

  1. dmidecode — project documentation and DMI reliability limits.
  2. Proxmox VE — node management documentation source.
  3. Proxmox VE — pvenode command reference.
  4. OpenSSH — ssh manual, including the -F option.