Superset
Superset connects through the SQLAlchemy dialect, so Thunderduck appears as a normal database — one connection, every catalog.
Add the database
Settings → Database Connections → + Database → Other, with:
thunderduck://:[email protected]Install thunderduck-sqlalchemy into the Superset image first — the driver has
to exist in the same environment as the web and worker processes.
Settings that matter
| Setting | Value | Why |
|---|---|---|
| Allow DML | off, unless you need it | Keeps dashboards read-only |
| Async query execution | on, if you have Celery | Long queries stop blocking the UI |
| Chart cache timeout | non-zero | Every chart refresh is a billed query otherwise |
Building charts
Use SQL Lab as you would anywhere, with the same three-part names. Datasets can be created from any query result, including one that joins across a Postgres catalog and an Iceberg catalog — from Superset's point of view it is one source.
Cost, deliberately
A dashboard that auto-refreshes turns into a query on every tick, and every query is priced. Set a cache timeout, and prefer datasets built on aggregates over raw tables for anything shown on a wall.