Quickstarts
Each quickstart is a complete scenario you can run end to end on a laptop, built on a Docker Compose stack from the repository's examples/ directory. Pick the one closest to what you want to do.
Route your first queryPostgres + Trino + QueryFlux + Studio. Send a query through the proxy and find it in query history.Run locally without DockerInstall the binary and query the embedded DuckDB engine — no containers, no other services.Build a multi-engine lakehouseTrino and StarRocks over Iceberg. Route Trino HTTP and MySQL wire clients to different engines.Monitor with Prometheus & GrafanaScrape QueryFlux metrics and watch queues, latency, and routing in Grafana dashboards.Query from an AI agent (MCP)Point Claude Code, Cursor, or MCP Inspector at QueryFlux and run SQL through an agent.Secure with Keycloak (OIDC)Authenticate clients with OIDC tokens and pass identity through to the backend.
Prerequisites
- Docker and Docker Compose
- Git, to clone the repository
- For Run locally without Docker: Rust and Python 3.10+ instead of Docker
git clone https://github.com/lakeops-org/queryflux.git
cd queryflux
Five-minute smoke test
If you only want to see QueryFlux answer a query:
cd examples/minimal-trino
docker compose up -d --wait
curl -X POST http://localhost:8080/v1/statement \
-H "X-Trino-User: dev" \
-d "SELECT 42"
Studio is at http://localhost:3000 (login admin / admin). Stop the stack with docker compose down -v.
Run one stack at a time
Several examples publish the same host ports (8080, 8081, 3000, 9000). Stop one stack before starting another, or change the published ports in its docker-compose.yml.
Ready to deploy?
The quickstarts use throwaway stacks. For a real deployment — Kubernetes with Helm, a standalone Docker container, or a native binary — see Installation & Deployment.