Skip to main content
Last updated on

What happened to Databricks Repos?

Repos are now called Git folders. Like Repos, Git folders let you sync workspace folders with remote Git repositories for version control.

What does this change mean for me?​

The core functionality hasn't changed, only the terminology. UI operations now refer to "Git folders" instead of "Repos".

Previously, you created a Git-backed folder by selecting New > Repo:

The "New" menu option used to refer to a "Repo"

Now, select New > Git folder.

The "New" menu option now asks you to create a "Git folder"

This change simplifies working with version-controlled folders:

  • Flexible folder organization: Create Git folders at any level of the workspace file tree. For example, /Workspace/Users/<user email>/level_1/level_2/level_3/<Git folder name>. Repos could only exist at a fixed level like /Workspace/Repos/<user email>/<Repo name>.

    note

    Git folders can contain asset types that Repos don't support, such as Databricks SQL assets and MLflow experiments. Databricks adds serialization support for additional assets over time.

  • Simplified UI: Work with Git directly in your workspace without navigating to a separate Repos area.

What changed?​

  • Git folders can exist outside the /Repos directory.
  • To create a Git folder, select New > Git folder. New Git folders appear under /Workspace/Users/<user-email>/.
  • Git folders can exist at any depth under /Workspace/Users/<user-email>/, and you can have multiple Git folders.
  • Git folders support asset types that Repos don't. Databricks adds serialization support for more asset types over time.
  • Git folders require a remote repository URL. Repos didn't have this requirement.

What happens to my current Repos?​

Repos are now integrated into the workspace UI under /Workspace/Repos instead of a separate top-level Repo node.

  • Existing /Repos paths continue to work. Both /Repos and /Workspace/Repos refer to the same folder, so paths in jobs, dbutils.notebook.run, and %run references don't need to change.
  • In rare cases, you might need a one-time workspace modification for this redirection to work. See References to workspace objects.

Databricks recommends creating new Git folders instead of Repos. Colocating Git repos with other workspace assets makes them more discoverable and easier to manage.

Git folder permissions​

Git folders use the same workspace folder permissions as other workspace folders. Most Git operations require the CAN_MANAGE permission.

Databricks Runtime version for running code in Git folders​

For consistent behavior between Git folders and legacy Repos, use Databricks Runtime 15.0 and above.

Current working directory (CWD) behavior​

Databricks Runtime 14.0 and above supports relative paths and provides a consistent CWD experience for all notebooks. Earlier Databricks Runtime versions might have inconsistent CWD behavior between Git folders and non-Git folders.

Python sys.path behavior​

Databricks Runtime 14.3 and above provides the same sys.path behavior in Git folders as in legacy Repos. Earlier versions don't automatically add the root repo directory to sys.path for Git folders. As a workaround, manually append the folder path to sys.path.

For examples, see Import Python and R modules.

Python library precedence​

Databricks Runtime 14.3 and above provides the same Python library precedence in Git folders as in legacy Repos.