Manage Postgres roles

Beta

Lakebase Postgres (Autoscaling Beta) is the next version of Lakebase, available for evaluation only. For production workloads, use Lakebase Public Preview. See choosing between versions to understand which version is right for you.

When you create a database project, Lakebase creates several Postgres roles in the project:

• A Postgres role for the project owner's Databricks identity (for example, user@databricks.com), which owns the default databricks_postgres database
• A databricks_superuser administrative role

The databricks_postgres database is created so you can connect and try out Lakebase immediately after project creation.

Several system-managed roles are also created. These are internal roles used by Databricks services for management, monitoring, and data operations.

Learn more: Pre-created roles and System roles

Create Postgres roles

Lakebase supports two types of Postgres roles for database access:

• Native Postgres password roles: Create these using the Lakebase UI or SQL. Use any valid role name with password authentication.
• OAuth roles for Databricks identities: Create these using the databricks_auth extension and SQL. Enables other Databricks identities to connect using OAuth tokens.

Learn more about authentication methods: Authentication

Create a native Postgres password role

You can create native Postgres password roles using either the Lakebase UI or standard SQL commands.

Using the UI:

1. In the Lakebase App, navigate to your branch overview page and then the branch's Roles & Databases tab.
2. Click Add role and specify a role name (any valid Postgres role name).

3. Click Create.
4. Copy the generated password and provide it securely to the user who will use this role.

The Lakebase UI generates a secure password automatically with 60-bit entropy. Roles created through the UI are automatically granted membership in the databricks_superuser role, which provides broad database privileges.

Using SQL:

You can also create native Postgres password roles using standard Postgres SQL commands:

SQL

CREATE ROLE role_name WITH LOGIN PASSWORD 'your_secure_password';

When creating roles with SQL, the password should have at least 12 characters with a mix of lowercase, uppercase, number, and symbol characters. User-defined passwords are validated at creation time to ensure 60-bit entropy.

Create an OAuth role for a Databricks identity using SQL

To allow other Databricks identities (users, service principals, or groups) to connect using OAuth tokens, you must create their Postgres roles using the databricks_auth extension.

Prerequisites:

• You must have CREATE and CREATE ROLE permissions on the database
• You must be authenticated as a Databricks identity with a valid OAuth token
• Native Postgres authenticated sessions cannot create OAuth roles

To create an OAuth role:

1. Create the databricks_auth extension. Each Postgres database must have its own extension.

   SQL

   CREATE EXTENSION IF NOT EXISTS databricks_auth;

2. Use the databricks_create_role function to create a Postgres role for the Databricks identity:

   SQL

   SELECT databricks_create_role('identity_name', 'identity_type');

   For a Databricks user:

   SQL

   SELECT databricks_create_role('myuser@databricks.com', 'USER');

   For a Databricks service principal:

   SQL

   SELECT databricks_create_role('8c01cfb1-62c9-4a09-88a8-e195f4b01b08', 'SERVICE_PRINCIPAL');

3. Grant database permissions to the newly created role.

The databricks_create_role() function creates a Postgres role with LOGIN permission only. After creating the role, you must grant the appropriate database privileges and permissions on the specific databases, schemas, or tables the user needs to access. Learn how: Manage permissions

note

Role names cannot exceed 63 characters, and some names are not permitted. Learn more: Manage roles

View Postgres roles

UI

To view all Postgres roles in your project, navigate to your branch's Roles & Databases tab in the Lakebase App. All roles created in the branch, with the exception of System roles, are listed, including native Postgres roles with password authentication.

PostgreSQL

View all roles with \du command:

You can view all Postgres roles, including system roles, using the \du meta-command from any Postgres client (such as psql) or the Lakebase SQL editor:

PostgreSQL

\du
                                      List of roles
          Role name          |                         Attributes
-----------------------------+------------------------------------------------------------
 cloud_admin                 | Superuser, Create role, Create DB, Replication, Bypass RLS
 my.user@databricks.com      | Create role, Create DB, Bypass RLS
 databricks_control_plane    | Superuser
 databricks_gateway          |
 databricks_monitor          |
 databricks_reader_12345     | Create role, Create DB, Replication, Bypass RLS
 databricks_replicator       | Replication
 databricks_superuser        | Create role, Create DB, Cannot login, Bypass RLS
 databricks_writer_12345     | Create role, Create DB, Replication, Bypass RLS

This displays all roles and their attributes (Superuser, Create role, Create DB, etc.).

Drop a Postgres role

You can drop both Databricks identity-based roles and native Postgres password roles.

UI

Deleting a role is a permanent action that cannot be undone. You cannot delete a role that owns a database. The database must be deleted before deleting the role that owns the database.

To delete any Postgres role using the UI:

1. Navigate to your branch's Roles & Databases tab in the Lakebase App.
2. Select Delete role from the role menu and confirm the deletion.

PostgreSQL

You can drop any Postgres role using standard Postgres commands. For details, see the PostgreSQL documentation on dropping roles.

Drop a role:

PostgreSQL

DROP ROLE role_name;

After a Databricks identity-based role is dropped, that identity can no longer authenticate to Postgres using OAuth tokens until a new role is created.

Pre-created roles

After a database project is created, Databricks automatically creates Postgres roles for project administration and getting started.

Role	Description	Inherited privileges
<project_owner_role>	The Databricks identity of the project creator (for example, my.user@databricks.com). This role owns the default databricks_postgres database and can log in and administer the project.	Member of databricks_superuser
databricks_superuser	An internal administrative role. Used to configure and manage access across the project. This role is granted broad privileges.	Inherits from pg_read_all_data, pg_write_all_data, and pg_monitor.

Learn more about these roles' specific capabilities and privileges: Pre-created role capabilities

System roles created by Databricks

Databricks creates the following system roles required for internal services. You can view these roles by issuing a \du command from psql or the Lakebase SQL Editor.

Role	Purpose
cloud_admin	Superuser role used for cloud infrastructure management
databricks_control_plane	Superuser role used by internal Databricks components for management operations
databricks_monitor	Used by internal metrics collection services
databricks_replicator	Used for database replication operations
databricks_writer_<dbid>	Per-database role used to create and manage synced tables
databricks_reader_<dbid>	Per-database role used to read tables registered in Unity Catalog
databricks_gateway	Used for internal connections for managed data serving services

To learn how roles, privileges, and role memberships work in Postgres, use the following resources in the Postgres documentation:

• Database Roles
• Privileges

Next steps

• Manage permissions - Grant database permissions to Postgres roles
• Connect to your database project - Connect using OAuth tokens or Postgres passwords
• Authentication - Understand how OAuth tokens and Postgres passwords work