Describe the bug
Starting with Databricks VS Code extension 2.14.0, eligible uv projects are routed to the managed Python environment setup flow, which invokes:
databricks environments setup-local
The command fetches runtime-specific constraints from raw.githubusercontent.com, for example:
https://raw.githubusercontent.com/databricks/environments/main/python/dbr/15.4.x-scala2.12/pyproject.toml
In corporate environments where raw.githubusercontent.com is blocked or restricted, this prevents Databricks Connect from being used in VS Code even when the project already has a valid local environment and a compatible databricks-connect installation.
This is a regression from the pre-2.14.0 workflow, where users could manage their Python environment themselves.
To Reproduce
-
Use Databricks VS Code extension version 2.14.0 or later.
-
Open a uv-suitable project with an existing .venv.
-
Ensure the environment already contains a compatible Databricks Connect version, for example:
databricks-connect>=15.4,<15.5
-
Use the extension to configure, run, or debug with Databricks Connect.
-
On a network where raw.githubusercontent.com is blocked, observe that the extension invokes databricks environments setup-local.
-
The setup fails in its fetch phase instead of using the existing interpreter/environment.
Actual behavior
The managed setup-local flow is selected for eligible uv projects and requires access to raw.githubusercontent.com.
Example failure:
Setup failed in the fetch phase (E_FETCH).
fetch constraints for dbr/15.4.x-scala2.12:
GET https://raw.githubusercontent.com/databricks/environments/main/python/dbr/15.4.x-scala2.12/pyproject.toml:
An existing connection was forcibly closed by the remote host.
Phases:
preflight=ok
resolve=ok
fetch=error
merge=pending
provision=pending
validate=pending
Expected behavior
Managed uv setup should remain available as a convenience, but it should not be mandatory.
The extension should provide a supported manual path, for example:
- A setting to disable managed Python environment setup.
- A choice between managed setup and an existing/manual interpreter.
- Automatic use of the selected interpreter when it contains a compatible
databricks-connect installation.
A valid local environment should not require access to raw.githubusercontent.com merely to use Databricks Connect from VS Code.
Impact
This blocks Databricks Connect workflows for corporate users whose security policy blocks GitHub Raw, despite their local Python environment already being correctly configured.
Downgrading the extension to 2.13.1 avoids the new flow, but is not a sustainable workaround because it prevents users from receiving later extension updates and fixes.
System information
- Databricks Extension Version: 2.14.x
- Compute target / DBR version: 15.4.x-scala2.12
Additional context
The release PR that made the setup flow generally available intentionally removed its feature flag: #2124
Describe the bug
Starting with Databricks VS Code extension 2.14.0, eligible
uvprojects are routed to the managed Python environment setup flow, which invokes:The command fetches runtime-specific constraints from
raw.githubusercontent.com, for example:In corporate environments where
raw.githubusercontent.comis blocked or restricted, this prevents Databricks Connect from being used in VS Code even when the project already has a valid local environment and a compatibledatabricks-connectinstallation.This is a regression from the pre-2.14.0 workflow, where users could manage their Python environment themselves.
To Reproduce
Use Databricks VS Code extension version 2.14.0 or later.
Open a
uv-suitable project with an existing.venv.Ensure the environment already contains a compatible Databricks Connect version, for example:
databricks-connect>=15.4,<15.5Use the extension to configure, run, or debug with Databricks Connect.
On a network where
raw.githubusercontent.comis blocked, observe that the extension invokesdatabricks environments setup-local.The setup fails in its fetch phase instead of using the existing interpreter/environment.
Actual behavior
The managed
setup-localflow is selected for eligibleuvprojects and requires access toraw.githubusercontent.com.Example failure:
Expected behavior
Managed
uvsetup should remain available as a convenience, but it should not be mandatory.The extension should provide a supported manual path, for example:
databricks-connectinstallation.A valid local environment should not require access to
raw.githubusercontent.commerely to use Databricks Connect from VS Code.Impact
This blocks Databricks Connect workflows for corporate users whose security policy blocks GitHub Raw, despite their local Python environment already being correctly configured.
Downgrading the extension to 2.13.1 avoids the new flow, but is not a sustainable workaround because it prevents users from receiving later extension updates and fixes.
System information
Additional context
The release PR that made the setup flow generally available intentionally removed its feature flag: #2124