Dagster has added beta support for the dbt Fusion engine in its dbt integration. dbt Fusion is a Rust-based standalone binary that compiles and parses dbt projects up to 30x faster than dbt Core, with additional features like state-aware caching, a SQL-aware language server, and improved column-level lineage. The integration required removing the hard dependency on dbt-core, adding compatibility shims for missing Python types, updating log parsing for Fusion's different event format, handling manifest differences (missing child_map field), and implementing engine-aware selection logic via CLI fallback. Setup requires installing the Fusion binary, optionally skipping dbt-core, and using the standard DbtCliResource configuration.

4m read timeFrom dagster.io
Post cover image
Table of contents
The technical details: How we added dbt Fusion support to Dagster
5 Impressions