Upgrade
Upgrade replaces the currently running agent with a newer version released. It is the standard path for rolling out bug fixes, security patches, or new features while preserving the device identity and configuration.
When it happens
- CLI (terminal):
nimbus system agent-upgrade - UI: Click the upgrade action from the robot overview page or fleet console when a new version is available.
How it works
- The agent requests the latest version available.
- Before changing anything, it creates a backup snapshot (
nimbus system agent-backup <auto-name>). - The updater stops the live agent, downloads the new binaries, replaces
nimbus/nimbusd, and restarts the service under the same configuration. - After restart, the agent reports the new version and resumes its regular responsibilities.
Why the backup matters
- Each upgrade uses the backup as the safety net; if the new binaries fail to run the revert flow can instantly restore the previous build without human intervention.
- Automatic backups are stored locally so they survive even if the upgrade tarball is removed or the device loses internet access mid-upgrade.
- Learn how the snapshots are stored, listed, and deleted on the Backup page.
- When you need to roll back after an upgrade failure, follow the steps on Restore (Rollback).
Notes
- Backup is always created before upgrade.
- Backup name is auto-generated (e.g.,
backup20260113125514). - Upgrade logic is controlled by shell scripts supplied by the backend service and runs independently of the core agent code.