This blog is written by Pascal Vogel, Senior Solutions Architect and Evan Pandya, Product Marketing Manager.
You can deploy Databricks Apps by pulling your app’s source code directly from a Git repository instead of requiring the source to live in the Databricks workspace files.
When deploying an app, you can now choose between two deployment sources:
- Workspace folder: use the Databricks CLI databricks sync command, Git folders, or Databricks Asset Bundles (DABs) to copy your source code to the workspace files.
- Git repository: configure a repo once at the app level, then deploy by selecting a Git reference (branch, tag, or commit SHA).
You can switch between workspace and Git sources at any time for the same app.
Git as the source of truth simplifies implementing software development best practices for your Databricks Apps deployments and offers a simplified path towards CI/CD.
Git-backed deployments are available in Beta and can be enabled as a workspace-level preview.
Let’s take a look at how Git-backed deployments work with a quick walkthrough.
Walkthrough: Deploying a Git-backed app
1. Push your app code to a Git repository
Start by pushing your app source code to a Git repository. This must include a valid entry point file (e.g., app.py or app.js) and other files as needed following the Databricks Apps deployment logic.
Git-backed deployments for Databricks Apps are supported by all major Git providers, including GitHub, GitLab, Bitbucket, and more.
In this example, we have a private Git repository on GitHub with a simple Dash app that we want to deploy to Databricks Apps.
2. Create a Git-backed app
Let’s create a new Databricks App backed by a Git repository:
- Navigate to Compute and choose the Apps tab.
- Create a custom app and provide a name and optional description.
- Choose Next: Configure Git
- Input your Git repository URL
- Select your Git provider
- Make any other configurations and Create the app.
The app compute resource will now start up. In the meantime, we can configure Git credentials if needed.
3. Configure Git credentials (for private repositories)
If you want to deploy your app from a private Git repository, the app’s service principal must have a Git credential configured.
On the app details page, note the Git credential required callout. Choose Configure Git credential to authenticate with your Git provider following the instructions in Connect your Git provider to Databricks.
In this example case of a GitHub repository, we can simply link our GitHub account with a few clicks.
4. Deploy from Git
Finally, let’s deploy the app:
- Go to your app detail page and choose Deploy and From Git.
- Enter a branch name, tag, or commit SHA as a Git reference and specify the Reference type.
- Choose Deploy.
That’s it! Databricks Apps will now pull the latest code from Git and deploy your app.
Conclusion
If you’re running Databricks Apps in production, deploying from Git is a natural default: it’s reproducible, auditable, and fits into CI/CD workflows. Configure the repo once, deploy from a branch for iteration, and use tags or commit SHAs when you want predictable, repeatable releases.
This also strengthens code governance: Git becomes the source of truth and, combined with Databricks Git security controls (like URL allowlists and access controls), it’s easier to standardize how apps are deployed across a workspace.
To get started with Git-backed app deployments, take a look at the updated Databricks Apps deployment documentation.
