» PostgreSQL Secrets Engine

Name: postgresql

The PostgreSQL secrets engine for Vault generates database credentials dynamically based on configured roles. This means that services that need to access a database no longer need to hardcode credentials: they can request them from Vault, and use Vault's leasing mechanism to more easily roll keys.

Additionally, it introduces a new ability: with every service accessing the database with unique credentials, it makes auditing much easier when questionable data access is discovered: you can track it down to the specific instance of a service based on the SQL username.

Vault makes use both of its own internal revocation system as well as the VALID UNTIL setting when creating PostgreSQL users to ensure that users become invalid within a reasonable time of the lease expiring.

This page will show a quick start for this secrets engine. For detailed documentation on every path, use vault path-help after mounting the secrets engine.

» Quick Start

The first step to using the PostgreSQL secrets engine is to mount it. Unlike the kv secrets engine, the postgresql secrets engine is not mounted by default.

$ vault secrets enable postgresql
Success! Enabled the postgresql secrets engine at: postgresql/

Next, Vault must be configured to connect to the PostgreSQL. This is done by writing either a PostgreSQL URL or PG connection string:

$ vault write postgresql/config/connection \

In this case, we've configured Vault with the user "root" and password "vaulttest", connecting to a PostgreSQL instance in AWS RDS. The "postgres" database name is being used. It is important that the Vault user have the GRANT OPTION privilege to manage users.

Optionally, we can configure the lease settings for credentials generated by Vault. This is done by writing to the config/lease key:

$ vault write postgresql/config/lease lease=1h lease_max=24h
Success! Data written to: postgresql/config/lease

This restricts each credential to being valid or leased for 1 hour at a time, with a maximum use period of 24 hours. This forces an application to renew their credentials at least hourly, and to recycle them once per day.

The next step is to configure a role. A role is a logical name that maps to a policy used to generated those credentials. For example, lets create a "readonly" role:

$ vault write postgresql/roles/readonly \
    sql="CREATE ROLE \"{{name}}\" WITH LOGIN PASSWORD '{{password}}' VALID UNTIL '{{expiration}}';
    GRANT SELECT ON ALL TABLES IN SCHEMA public TO \"{{name}}\";"
Success! Data written to: postgresql/roles/readonly

By writing to the roles/readonly path we are defining the readonly role. This role will be created by evaluating the given sql statements. By default, the {{name}}, {{password}} and {{expiration}} fields will be populated by Vault with dynamically generated values. This SQL statement is creating the named user, and then granting it SELECT or read-only privileges to tables in the database. More complex GRANT queries can be used to customize the privileges of the role. See the PostgreSQL manual for more information.

To generate a new set of credentials, we simply read from that role: Vault is now configured to create and manage credentials for Postgres!

$ vault read postgresql/creds/readonly
Key               Value
---               -----
lease_id          postgresql/creds/readonly/c888a097-b0e2-26a8-b306-fc7c84b98f07
lease_duration    3600
password          34205e88-0de1-68b7-6267-72d8e32c5d3d
username          root-1430162075-7887

By reading from the creds/readonly path, Vault has generated a new set of credentials using the readonly role configuration. Here we see the dynamically generated username and password, along with a one hour lease.

Using ACLs, it is possible to restrict using the postgresql secrets engine such that trusted operators can manage the role definitions, and both users and applications are restricted in the credentials they are allowed to read.

If you get stuck at any time, simply run vault path-help postgresql or with a subpath for interactive help output.


The PostgreSQL secrets engine has a full HTTP API. Please see the PostgreSQL secrets engine API for more details.