Proxmox VE
pvesh GET: nodes, QEMU/LXC guests, NICs and resources
Proxmox inventory.Drift control with NetBox and Zabbix.
Designed and deployed an operational inventory workflow, from collecting hypervisor and VM configurations to reconciling them against NetBox and monitoring daily drift.
READ-ONLYScheduled discovery and audits without automatic infrastructure changes
0 DRIFTDifferences found during acceptance testing
RU / ENPublic, sanitized code and documentation
01 / CHALLENGE
A reliable inventory, not another spreadsheet.
Virtual machines, network adapters and cluster settings change continuously. Manual records become stale, while bulk synchronization mistakes can affect production. The goal was to maintain a usable NetBox inventory and detect changes without automatically pushing configuration updates.
02 / ARCHITECTURE
Separate discovery, inventory and verification.
Each component has a defined role, with read-only API access separate from privileged write operations.
Python
Normalized inventory, completeness checks and private JSON snapshots
NetBox
DCIM and Virtualization: hosts, VMs and virtual NICs
systemd
Separate daily physical-host and guest drift checks
Zabbix
CLEAN / DRIFT / ERROR, difference count and audit timestamp
03 / IMPLEMENTATION
From scripts to an operational workflow.
Host and VM discovery
Collect hypervisor hardware, QEMU/LXC guests, assigned CPU/RAM and virtual networking. Reconcile records against existing NetBox objects using stable identifiers.
Guarded import
Reviewable preview and dry-run, expected-count checks, a PostgreSQL backup, explicit confirmation and transactional interface creation. No automatic remediation.
Intentional exceptions
A stopped VM template may share a MAC with a live guest. The NIC is inventoried without assigning the conflicting address, and the exception is verified by the audit.
Drift monitoring
Separate systemd timers and report paths for hosts and guests, sanitized numeric export via Zabbix Agent, and triggers for drift, execution errors and stale data.
04 / OUTCOME
Auditable inventory rather than manual tracking.
The production NetBox inventory was verified against Proxmox for hosts, guests and network interfaces. End-to-end validation included Zabbix Server polling the exported values. The system detects drift; it is not a bidirectional infrastructure orchestrator.
- Independent daily hardware and QEMU/LXC drift checks
- Distinct CLEAN, DRIFT and ERROR states
- Snapshot freshness and exception validation
- Restricted file permissions, TLS and scoped read-only API credentials
- Python tests and GitHub Actions CI checks
05 / SOURCE CODE
Engineering decisions backed by code.
The public repository includes collectors, auditors, systemd definitions, Zabbix integration, tests and operational documentation. Production addresses, internal hostnames, VMIDs and MACs are excluded.
Scope: inventory and drift detection. NetBox does not directly manage Proxmox configuration, and drift never triggers automatic changes. The code requires environment-specific configuration and validation before use in production.