v1.7.3
What's Changed
The big change in this release is that we fixed the issue where every single dbt action initiated a new connection to Databricks. We will now reuse a connection if there is a thread-local connection that matches the compute the user has selected.
This change will be most apparent if your dbt operations are very short lived, such as tests against a small table, as there is now less time spent in connection negotiation; for longer operations, the time spent in computing and transmitting the result set is more significant than the time spent on connecting.
If for some unforeseen reason this change negatively impacts performance:
a.) You can turn it off by setting the DBT_DATABRICKS_LONG_SESSIONS environment variable to false.
b.) Please file an issue so we can investigate.
Fixes
- Allow session re-use by @rcypher-databricks in #517
- Fixing list_schemas so that we stop trying to create schemas that exist by @benc-db in #529
Under the Hood
- Refactor macro tests so that we can move macros by @benc-db in #524
- Updating Python Functional Tests by @benc-db in #526
- Refactoring to align with dbt-core organization: Part I by @benc-db in #525
Full Changelog: v1.7.2...v1.7.3
