databricks-sql-python icon indicating copy to clipboard operation
databricks-sql-python copied to clipboard

User/interactive Azure AD authentication

Open inigohidalgo opened this issue 4 months ago • 0 comments

#621 by @jprakash-db added a great feature that we will be testing ASAP. Since Entra ID authentication seems to be getting upgraded, I wanted to ask about interactive login.

Using databricks-sdk I can perform this user authentication by doing (from the terminal)

az login --use-device-code
...
User is logged in

then in python

from databricks.sdk import WorkspaceClient

w = WorkspaceClient(host="https://adb-XXXXXX.azuredatabricks.net/")

Would the newly-added Entra ID M2M method be extendable for U2M workflows?

Otherwise, is there an alternative implementation I could use?

This is what I'm currently doing to manually mint a token in the user interactive workflow currently:

from azure.identity import DefaultAzureCredential
from databricks import sql as dbx_sql

_DATABRICKS_SCOPE = "2ff814a6-3304-4ab8-85cb-cd0e6f879c1d/.default"

DATABRICKS_CLUSTER_HOST = "adb-XXX.azuredatabricks.net"
DATABRICKS_CLUSTER_HTTP_PATH = "/sql/1.0/warehouses/XXX"

cred = DefaultAzureCredential()
token = cred.get_token(_DATABRICKS_SCOPE)

with dbx_sql.connect(
    server_hostname=DATABRICKS_CLUSTER_HOST,
    http_path=DATABRICKS_CLUSTER_HTTP_PATH,
    access_token=token.token,
) as conn, conn.cursor() as c:
        c.execute("SELECT current_user(), current_catalog(), current_schema();")
        print(c.fetchall())

Thanks!

inigohidalgo avatar Sep 01 '25 15:09 inigohidalgo