Operations

Operational tasks protect an existing deployment and help explain its behavior. Before changing storage or software, identify the exact data root, database backend, local artifact files, plugins, and release version in use.

Select an operations guide

• Backup and restore : plan a consistent recovery set across the database and local files, including when PostgreSQL is used.
• Upgrade : check the release compatibility boundary, preserve a recoverable pre-upgrade state, and avoid unsupported legacy layouts.
• Storage : understand the data-root layout, retained logs and observations, and the difference between reported usage and storage capacity.
• Diagnostics : collect startup or request evidence without exposing tokens, provider credentials, or sensitive interaction data.

Preserve a recoverable state

The database is not necessarily the whole deployment. Local artifacts, imported vendor components, configuration, and other data-root state may remain outside a remote PostgreSQL database. Treat diagnostics and debug exports as sensitive; ordinary request observations can include content as well. Share the minimum information needed and review it before sending it outside a trusted environment.

Check results, not just process state

A process that starts or returns a health response may still be in first-time setup, unable to connect to its configured database, or not ready to serve a model. Verify the relevant readiness state and an actual client request after operational changes. Follow Troubleshooting when a specific symptom remains.