Skip to content

03-storage.md

Latest commit

 

History

History
57 lines (38 loc) · 1.91 KB

03-storage.md

File metadata and controls

57 lines (38 loc) · 1.91 KB

Storage

This document describes how Geti Instant Learn stores and manages persistent data, including the file system layout, database configuration, and volume mounting for Docker deployments.

Overview

Geti Instant Learn uses a SQLite database for configuration persistence and a file-based structure for media assets (datasets, prompts, thumbnails).

File System Layout

data/                              # DB_DATA_DIR
├── instant_learn.db                 # SQLite database
└── templates/
    └── datasets/
        └── aquarium/              # Sample dataset (example)

logs/                              # LOGS_DIR
└── instant-learn-backend.log        # Application log

Database

SQLite Configuration

Geti Instant Learn uses SQLite as its embedded database for storing:

  • Project configurations
  • Source settings (cameras, video files, image folders)
  • Sink configurations (MQTT, custom outputs)
  • Model selections and parameters
  • Prompt definitions (visual and text prompts)
  • Label schemas

Schema Management

Database migrations are managed with Alembic. Migration scripts are located in application/backend/app/domain/alembic/versions/.

Docker Volume Mounting

Container Mount Points

Container Path Purpose
/instant_learn/data Database and user data
/instant_learn/logs Application logs

Note: The container runs as non-root user (UID 10001). For bind mounts, ensure host directories have correct permissions.

Configuration Reference

Environment Variables

Variable Description
DB_DATA_DIR Database and data directory
LOGS_DIR Log files directory