PostgreSQL Automatic User Provisioning

Teleport can automatically create users in your database, removing the need for creating individual user accounts in advance or using the same set of shared database accounts for all users.

Prerequisites

• Teleport cluster v13.1 or above with a configured self-hosted PostgreSQL or RDS PostgreSQL database. To configure permissions for database objects like tables, your cluster must be on version v15.2 or above.
• Ability to connect to and create user accounts in the target database.

Step 1/3. Configure database admin

Teleport should be able to connect to the database as a user that can create other users and assign them roles. We recommend creating a separate user designated specifically for Teleport automatic user provisioning. Let's call it teleport-admin.

Teleport will use the same authentication mechanism when connecting as an admin user as for regular user connections: X.509 for self-hosted databases and AWS IAM for RDS. The admin user must have privileges within the database to create users and grant them privileges, either for roles or concrete database objects.

CREATE USER "teleport-admin" login createrole;
GRANT rds_iam TO "teleport-admin" WITH ADMIN OPTION;

CREATE USER "teleport-admin" login createrole;

hostssl all             all             ::/0                    cert
hostssl all             all             0.0.0.0/0               cert

GRANT ALL PRIVILEGES ON ALL TABLES IN SCHEMA schema1, schema2, schema3 TO "teleport-admin";

Users created by Teleport will be placed in the teleport-auto-user group in the database, which will be created automatically if it doesn't exist.

Next, configure the database admin user in the Teleport database configuration:

kind: db
version: v3
metadata:
  name: example
spec:
  protocol: "postgres"
  uri: "localhost:5432"
  admin_user:
    name: "teleport-admin"
    # Optional default database the admin user logs into. Default is
    # the same database that the user is accessing, if not specified.
    # default_database: teleport

This example assumes that you have configured the database as a dynamic resource. If you have configured your database using a static Teleport Database Service configuration, edit the entry in your db_service.databases configuration.

For auto-discovered cloud databases, the name of the admin user is taken from the teleport.dev/db-admin label, and the default database is taken from the teleport.dev/db-admin-default-database label.

GRANT CREATE ON SCHEMA public TO "teleport-admin";

Step 2/3. Configure a Teleport role

Database permissions are associated with a Teleport role, which can either allocate predefined database roles (configured in each database) or define specific database object permissions directly. Teleport grants these permissions for the duration of the connection.

With automatic user provisioning, users always connect to the database with their Teleport username so the db_users role field is ignored for roles that have database user provisioning enabled.

The available provisioning modes are:

• off: Disables user provisioning.

• keep: Enables user provisioning and disables users at session end. The user will be stripped of all roles and the user account will be locked.

• best_effort_drop: Enables user provisioning and, when the session ends, drops the user if no resources depend on it. In cases where any resource depends on the user, it falls back to disabling the user, mirroring the behavior of keep mode.

kind: role
version: v7
metadata:
  name: auto-db-users
spec:
  options:
    # create_db_user_mode enables automatic user provisioning for matching databases
    create_db_user_mode: keep
  allow:
    db_labels:
      "*": "*"
    db_names:
    - "*"
    # db_roles is a list of roles the database user will be assigned
    db_roles:
    - reader
    - "{{internal.db_roles}}"
    - "{{external.db_roles}}"

kind: role
metadata:
  name: read_all_tables
spec:
  options:
    # create_db_user_mode enables automatic user provisioning for matching databases
    create_db_user_mode: keep
  allow:
    db_labels:
      '*': '*'
    db_names:
    - '*'
    db_permissions:
    # grant `SELECT` on all tables in `public` schema.
    - match:
        # object labels to match
        object_kind: table
        schema: public
      permissions:
      - SELECT
version: v7

Users created within the database will:

• Have the same username as the authenticated Teleport user.
• Be a part of the teleport-auto-user role.
• Will be assigned permissions according to the chosen mechanism.

Note that in case of a name conflict where a user with the same name already exists in the database and is not managed by Teleport (i.e. not assigned the teleport-auto-user role), the connection will be aborted.

Step 3/3. Connect to the database

Now, log into your Teleport cluster and connect to the database:

tsh login --proxy=teleport.example.com
tsh db connect --db-name <database> example

To view the list of database roles that are allowed for each database, you can use the command tsh db ls -v. By default, all database roles will be assigned to your auto-provisioned database user. You can optionally select a subset of the database roles with --db-roles:

tsh db connect --db-name <database> --db-roles reader example

Troubleshooting

Permission denied for schema public error

PostgreSQL 15 revokes the CREATE permission from all users except a database owner from the public (or default) schema.

Grant the admin user CREATE privilege so the admin user can create procedures:

GRANT CREATE ON SCHEMA public TO "teleport-admin";

If admin_user.default_database is specified, the CREATE privilege is only required for the database specified in the default_database. Otherwise, you have to repeat the privilege grant for every database Teleport will access.

User does not have CONNECT privilege error

You may encounter the following error when the admin user or the roles assigned to the auto-provisioned user do not have permission to connect to the target database:

tsh db connect --db-name <database> example
...FATAL: permission denied for database "<database>"DETAIL: User does not have CONNECT privilege.

Make sure CONNECT is granted to the admin user and the respective roles:

GRANT CONNECT ON DATABASE <database> to "teleport-admin";
GRANT CONNECT ON DATABASE <database> to "reader";

Cannot execute in a read-only transaction error

You may encounter the following error when connecting to an Amazon RDS Aurora reader endpoint:

tsh db connect --db-name <database> example
psql: error: connection to server at "localhost" (::1), port 12345 failed: Connection refused...ERROR: cannot execute CREATE ROLE in a read-only transaction (SQLSTATE 25006)

Database auto-user provisioning is not compatible with RDS Aurora reader endpoints. Please use auto-user provisioning on the primary endpoints.

No schema has been selected error

You may encounter the following error when connecting if the admin user does not have USAGE permission on the schema:

tsh db connect --db-name <database> example
psql: error: connection to server at "localhost" (::1), port 12345 failed: Connection refused...ERROR: no schema has been selected to create in (SQLSTATE 3F000)

To fix this, make sure that the admin user is granted USAGE and CREATE on schema public in the target database:

GRANT USAGE ON SCHEMA public TO "teleport-admin";
GRANT CREATE ON SCHEMA public TO "teleport-admin";

Permission denied to grant role "rds_iam"

You may encounter the following error when connecting to an RDS database:

tsh db connect --db-name <database> example
psql: error: connection to server at "localhost" (::1), port 12345 failed: Connection refused...ERROR: permission denied to grant role "rds_iam" (SQLSTATE 42501)

This happens when the admin user does not have permission to grant the "rds_iam" role to other users. To fix this, grant the "rds_iam" role with the ADMIN option to the admin user:

GRANT rds_iam TO "teleport-admin" WITH ADMIN OPTION;

Use your mapped remote username error

You may encounter the following error when connecting to a database in a remote cluster:

> tsh db connect --db-name <database> exampleERROR: please use your mapped remote username ("remote-<your-teleport-username>-<root-cluster-name>") to connect instead of "<database-user>"

When you access resources in a remote cluster, the remote cluster will receive the name remote-<your-teleport-username>-<root-cluster-name> from the local cluster. This is to prevent any naming collisions with users in the remote cluster. Please use the username from the error message as the database username for when connecting through tsh or GUI clients.

Unable to cancel a query

If you use a PostgreSQL cli client like psql, and you try to cancel a query with ctrl+c, but it doesn't cancel the query, then you need to connect using a tsh local proxy instead. When psql cancels a query, it establishes a new connection without TLS certificates, however Teleport requires TLS certificates not only for authentication, but also to route database connections.

If you enable TLS Routing in Teleport then tsh db connect will automatically start a local proxy for every connection. Alternatively, you can connect via Teleport Connect which also uses a local proxy. Otherwise, you need to start a tsh local proxy manually using tsh proxy db and connect via the local proxy.

If you have already started a long-running query in a psql session that you cannot cancel with ctrl+c, you can start a new client session to cancel that query manually:

First, find the query's process identifier (PID):

SELECT pid,usename,backend_start,query FROM pg_stat_activity WHERE state = 'active';

Next, gracefully cancel the query using its PID. This will send a SIGINT signal to the postgres backend process for that query:

SELECT pg_cancel_backend(<PID>);

You should always try to gracefully terminate a query first, but if graceful cancellation is taking too long, then you can forcefully terminate the query instead. This will send a SIGTERM signal to the postgres backend process for that query:

SELECT pg_terminate_backend(<PID>);

See the PostgreSQL documentation on admin functions for more information about the pg_cancel_backend and pg_terminate_backend functions.

SSL SYSCALL error

You may encounter the following error when your local psql is not compatible with newer versions of OpenSSL:

tsh db connect --db-user postgres --db-name postgres postgres
psql: error: connection to server at "localhost" (::1), port 12345 failed: Connection refused    Is the server running on that host and accepting TCP/IP connections?connection to server at "localhost" (127.0.0.1), port 12345 failed: SSL SYSCALL error: Undefined error: 0

Please upgrade your local psql to the latest version.

Amazon RDS Blue/Green deployment enters a state of "Replication degraded"

Amazon RDS Blue/Green deployment may enter a state of "Replication degraded" when auto-user provisioning is used to connect to the database.

This occurs due to a limitation in PostgreSQL logical replication for Blue/Green deployments as Data Definition Language (DDL) statements cannot be replicated. Consequently, Amazon RDS will enter the "Replication degraded" state when a DDL change is detected.

It is recommended to disable database auto-user provisioning before starting the Blue/Green deployment.

If the database is auto-discovered by Teleport, you can temporarily remove the teleport.dev/db-admin AWS resource tag. For a database registered using either static config or a dynamic db resource, you can temporarily remove the admin_user setting.

Once auto-user provisioning is disabled, you can still connect as the database admin user through Teleport.

Next steps

• Connect using your GUI database client.
• Learn about role templating.
• Read automatic user provisioning RFD.
• Read database permission management RFD.
• The internal.db_roles traits we illustrated in this guide are replaced with values from the Teleport local user database. For full details on how variable expansion works in Teleport roles, see the Teleport Access Controls Reference.