NVIDIA UFM Enterprise User Manual v6.24.1

Appendix - UFM Versions Manager

Note
  • Supports UFM 6.22.2 and later.

  • Supports UFM deployments running in containers with Podman or Docker.

  • Supports both Standalone (SA) and High Availability (HA) environments.

  • UFM Infra mode is not supported by this tool.

UFM Versions Manager is a command-line tool for managing UFM system backups, configurations, and migrations. It provides comprehensive backup and restore capabilities for UFM systems running in Podman containers, with full support for both Standalone (SA) and High Availability (HA) environments.

The tool provides four core capabilities:

  • Configuration Snapshots - Fast config-only backups (up to 10 snapshots with rotation)

  • Full System Backup/Restore - Complete disaster recovery including images and configurations (up to 5 backups with rotation when using the default backup location)

  • Migration - Move UFM to new hardware (SA or HA environments)

  • Upgrade - Upgrade UFM to new versions with automatic backup and version validation

  • Automatic environment detection (SA vs HA)

  • Rootless Podman support

  • Version compatibility checking

  • HA cluster support with master/standby coordination

  • Dry-run mode for all operations (--dry-run)

  • Parallel processing for faster backups (--max-workers)

  • Comprehensive logging to stdout and syslog

The UFM Version Manager tool is deployed as a separated package which can be downloaded from NVIDIA's Licensing Portal .

System Requirements

  • Python 3.10 or higher (up to Python 3.12, including)

  • Root access

  • Podman container runtime

  • UFM in Docker (only Docker/Podman UFM deployments supported)

  • Sufficient disk space:

    • Config snapshots: ~10 MB per snapshot

    • Full system backups: 3-10 GB, depending on the number of plugins

HA Environment Requirements

  • HA version 2.1.0 or higher (for migration operations only)

  • In HA setups:

    • Upgrade operation can only run from the Standby node.

    • Other commands must run from the master node

    • SSH access to the standby node

  • Ubuntu 22.04 / 24.04

  • RHEL 8 / 9

  • Oracle Linux 8 / 9 / 10

  • UFM Enterprise XDR Appliance (Gen 3.5) (see limitations)

Log Locations

Log

Location

stdout (Console)

  • All operations logged to console output in real-time

  • Shows INFO, WARNING, and ERROR messages

Syslog

  • WARNING and ERROR messages logged to system syslog

  • View with: journalctl | grep ufm_versions_mgr


  • Default: INFO level (shows operation progress)

  • Verbose mode: DEBUG level (use --verbose flag for detailed output)

Viewing Logs

Copy
Copied!
            

# View syslog entries journalctl | grep ufm_versions_mgr


Limitation

Description

UFM Infra Not Supported

If gv.cfg contains [UFMInfra] enabled=true , the tool exits and instructs to disable UFMInfra.

HA Configurations Not Backed Up

HA cluster configuration itself is not saved/restored by this tool.

Full Backup Retention

Using the default backup location, full system backups maintain up to 5 backups with rotation ( /opt/ufm/backup/downgrade/1..5 ). With a custom --backup-dir , rotation is not applied (single location).

Strict Version Compatibility

Config snapshots can only be restored when UFM version and plugin versions match.

HA Full Backup on Local Storage

Full backups are stored on the master node's local storage (`/opt/ufm/backup/downgrade/`), NOT on DRBD shared storage. After HA failover, the new master cannot access backups created on the old master. User must manually copy the backup directory from old master to new master if restore is needed. Configuration snapshots are unaffected (stored on shared storage

Root Access Required

The tool must run as root user

No Concurrent Operations

Only one instance can run at a time (operation lock).

Upgrade Not Available on Appliance

The upgrade command is blocked on UFM Appliance (detected via /etc/ufm-release).

Operation

Command

Downtime

Config snapshot

ufm_versions_mgr backup --config

No

Config restore

ufm_versions_mgr restore --config --snapshot-number 1

Yes

List config snapshots

ufm_versions_mgr backup --config --list

Yes

Config version check

ufm_versions_mgr restore --config --check-version --snapshot-number 1

No

Full backup

ufm_versions_mgr backup

No

List full system backups

ufm_versions_mgr backup --list

Yes

Full restore (SA)

ufm_versions_mgr restore

Yes

Full restore (HA)

ufm_versions_mgr restore

Yes

Full restore (select backup slot)

ufm_versions_mgr restore --backup-number N

Yes

SA migration

ufm_versions_mgr restore --migrate-sa --target-ip X.X.X.X

No

HA migration

ufm_versions_mgr restore --migrate-ha --target-ip X.X.X.X --hacluster-password PWD

Yes

Upgrade (SA)

ufm_versions_mgr upgrade FILE.gz

Yes

Upgrade (HA)

ufm_versions_mgr upgrade FILE.gz

No

Note

Full restore notes (tokens/users):

By default, full restore preserves tokens and users from the current UFM.

To restore them from the backup instead, use:

--restore-tokens

--restore-users

Note

ufm_versions_mgr backup --list shows master backups and attempts to include standby backups too (requires SSH trust to standby; otherwise it warns and shows master only).

Description

Command

General help

ufm_versions_mgr --help

Backup command help

ufm_versions_mgr backup --help

Restore command help

ufm_versions_mgr restore --help

Upgrade command help

ufm_versions_mgr upgrade --help

Show version

ufm_versions_mgr --version

Note

Use --verbose for detailed output during operations.

Use --dry-run to preview any operation before executing.

Config snapshots: /opt/ufm/files/backup/configurations/

Full system backups: /opt/ufm/backup/downgrade/

© Copyright 2026, NVIDIA. Last updated on Feb 20, 2026