Lakevision is a data lake observability tool built around the Apache Iceberg format. It provides a FastAPI backend with a SvelteKit frontend to enumerate namespaces and tables in your Iceberg catalog, display each table’s schema, properties, partitions, snapshots, and sample data, and offer insights into data changes over time. It also supports optional authentication via OIDC/OAuth and pluggable authorization. You can run Lakevision via Docker using the provided image, then access the UI at the host port you map (default work path is the UI on port 8081).

Once running, you can browse namespaces and tables, inspect individual table schemas and properties, review partition specs and sort orders, view recent snapshots, and see a graphical summary of record additions over time. If you enable authentication, you’ll be prompted to sign in, after which you’ll gain access control to datasets. There is also an optional in-app “Chat with Lakehouse” capability for quick data context queries.

In short, Lakevision acts as a centralized, UI-driven explorer for your Iceberg-based Lakehouse, helping data teams understand catalog contents, track changes, and assess data layout with minimal setup.

Prerequisites:

• Docker or a Python/Node development environment (as you prefer to run Lakevision in Docker, the Docker route is recommended for quick start)
• An Iceberg-compatible catalog accessible from the container (or a sample catalog if you choose to enable it)

Option A - Run Lakevision with Docker (recommended):

1. Build or pull the Lakevision image (example tag used in the README): docker build -t lakevision:1.0 .

   or if using a prebuilt image from a registry: docker pull lakevision:1.0

2. Prepare environment (copy template and edit as needed): cp my.env .env

   edit .env to configure Iceberg catalog, authentication, and optional cloud settings

3. Run the Lakevision container (exposes UI on port 8081 by default): docker run --env-file .env -p 8081:8081 lakevision:1.0 /app/start.sh

4. (Optional) Run the health worker container if health checks are enabled: docker run --env-file .env lakevision:1.0 /app/worker.sh

5. Open http://localhost:8081 in your browser to access the Lakevision UI.

Option B - Run locally (be and fe):

1. Prerequisites: Python 3.10+, Node.js 18+

2. Backend (be): cd be python -m venv .venv source .venv/bin/activate pip install -r requirements.txt PYTHONPATH=app uvicorn app.api:app --reload --port 8000

3. Frontend (fe): cd fe npm install npm run dev -- --port 8081

4. Ensure you have an Iceberg catalog accessible from the environment and the .env/OPENID settings if using Auth.

Prerequisites overview:

• Docker or a Python/Node dev environment
• Access to an Iceberg catalog (local or remote)
• Optional: OpenID Connect provider configuration for authentication

Tips and common issues:

• If you’re using Docker, make sure your .env file is correctly populated with Iceberg catalog settings and credentials required by Lakevision.
• The UI runs behind Nginx in the example Docker setup; ensure port mappings (e.g., 8081) don’t conflict with existing services.
• For development, you can enable the in-memory sample catalog by rebuilding the image with ENABLE_SAMPLE_CATALOG=true in the Docker build context as noted in the README.
• Environment variables that start with PUBLIC_ or AUTH_ are used for frontend and authentication configuration; ensure they are exposed to the frontend build when running locally via the /fe path.
• If you enable Authz, you’ll need to implement an Authz class per instructions and set PUBLIC_AUTH_ENABLED and related PUBLIC_OPENID_* variables.
• If you run the backend and frontend separately (manual setup), ensure PYTHONPATH and Vite environment variables are available to the frontend build (use make prepare-fe-env as described in the README).

Configuration options:

• Iceberg catalog URI and warehouse path in your .env
• Authentication settings (OIDC provider URL, client IDs, and redirect URI)
• Optional cloud storage settings for S3/GCP in your catalog configuration