Connector
Databricks
Read and write Databricks SQL through a SQL warehouse. Results stream back columnar, with no driver to install. Press Refresh columns on a node to read a query's columns; it starts a stopped warehouse, so it never happens on its own.
Settings
What the connection form asks for. Every connection has a Staging side, which it needs, and an optional Production side with values of its own; both ask for the same settings.
| Workspace Host | e.g. dbc-1234abcd-5678.cloud.databricks.com (with or without https://). |
| SQL Warehouse ID | From the warehouse's Connection details tab. Pasting the whole HTTP path (/sql/1.0/warehouses/…) works too. |
| Default catalog | Optional default catalog (e.g. main); a node can still address any catalog. |
| Default schema | Optional default schema (e.g. default). |
| Access Token | A personal access token for the workspace (Settings → Developer → Access tokens). Writing needs ALL API scopes, not just BI Tools: loading a table stages a file in a Unity Catalog volume, and a BI-Tools-scoped token can query but not upload. Stored encrypted, and never shown again after you save it. |
Testing the connection
Test runs SELECT 1 on the SQL warehouse.
Good to know
- A token scoped to BI Tools passes Test and still cannot load a table. Loading stages a file in a Unity Catalog volume, which needs a token with all API scopes.
- A node reads its columns only when you press Refresh columns, because that starts a stopped warehouse.