Using OpenID Connect (OIDC) and Keycloak to Centralize Authorization

To complete this guide, you need:

In this example, we build a very simple microservice which offers two endpoints:

These endpoints are protected and can only be accessed if a client is sending a bearer token along with the request, which must be valid (e.g.: signature, expiration and audience) and trusted by the microservice.

The bearer token is issued by a Keycloak Server and represents the subject to which the token was issued for. For being an OAuth 2.0 Authorization Server, the token also references the client acting on behalf of the user.

The /api/users/me endpoint can be accessed by any user with a valid token. As a response, it returns a JSON document with details about the user where these details are obtained from the information carried on the token. This endpoint is protected with RBAC (Role-Based Access Control) and only users granted with the user role can access this endpoint.

The /api/admin endpoint is protected with RBAC (Role-Based Access Control) and only users granted with the admin role can access it.

This is a very simple example using RBAC policies to govern access to your resources. However, Keycloak supports other types of policies that you can use to perform even more fine-grained access control. By using this example, you’ll see that your application is completely decoupled from your authorization policies with enforcement being purely based on the accessed resource.

We recommend that you follow the instructions in the next sections and create the application step by step. However, you can go right to the completed example.

Clone the Git repository: git clone https://github.com/quarkusio/quarkus-quickstarts.git, or download an archive.

The solution is located in the security-keycloak-authorization-quickstart directory.

First, we need a new project. Create a new project with the following command:

For Windows users:

This command generates a project, importing the keycloak-authorization extension which is an implementation of a Keycloak Adapter for Quarkus applications and provides all the necessary capabilities to integrate with a Keycloak Server and perform bearer token authorization.

If you already have your Quarkus project configured, you can add the oidc and keycloak-authorization extensions to your project by running the following command in your project base directory:

This will add the following to your build file:

Let’s start by implementing the /api/users/me endpoint. As you can see from the source code below it is just a regular Jakarta REST resource:

The source code for the /api/admin endpoint is also very simple:

Note that we did not define any annotation such as @RolesAllowed to explicitly enforce access to a resource. The extension will be responsible to map the URIs of the protected resources you have in Keycloak and evaluate the permissions accordingly, granting or denying access depending on the permissions that will be granted by Keycloak.

To start a Keycloak Server you can use Docker and just run the following command:

where keycloak.version should be set to 17.0.0 or higher.

You should be able to access your Keycloak Server at localhost:8543.

Log in as the admin user to access the Keycloak Administration Console. Username should be admin and password admin.

Import the realm configuration file to create a new realm. For more details, see the Keycloak documentation about how to create a new realm.

After importing the realm you can see the resource permissions:

It explains why the endpoint has no @RolesAllowed annotations - the resource access permissions are set directly in Keycloak.

To run the application in dev mode, use:

Dev Services for Keycloak will launch a Keycloak container and import a quarkus-realm.json.

Open a Dev UI available at /q/dev-v1 and click on a Provider: Keycloak link in an OpenID Connect Dev UI card.

You will be asked to log in into a Single Page Application provided by OpenID Connect Dev UI:

When you’re done playing with the dev mode" you can run it as a standard Java application.

First compile it:

Then run it:

This same demo can be compiled into native code: no modifications required.

This implies that you no longer need to install a JVM on your production environment, as the runtime technology is included in the produced binary, and optimized to run with minimal resource overhead.

Compilation will take a bit longer, so this step is disabled by default; let’s build again by enabling the native profile:

After getting a cup of coffee, you’ll be able to run this binary directly:

See Running the Application in Dev mode section above about testing your application in a dev mode.

You can test the application launched in JVM or Native modes with curl.

The application is using bearer token authorization and the first thing to do is obtain an access token from the Keycloak Server in order to access the application resources:

The example above obtains an access token for user alice.

Any user is allowed to access the http://localhost:8080/api/users/me endpoint which basically returns a JSON payload with details about the user.

The http://localhost:8080/api/admin endpoint can only be accessed by users with the admin role. If you try to access this endpoint with the previously issued access token, you should get a 403 response from the server.

In order to access the admin endpoint you should obtain a token for the admin user:

In some cases, you may want to use the Keycloak Authorization Client Java API to perform specific operations like managing resources and obtaining permissions directly from Keycloak. For that, you can inject a AuthzClient instance into your beans as follows:

Note: If you want to use the AuthzClient directly make sure to set quarkus.keycloak.policy-enforcer.enable=true otherwise there is no Bean available for injection.

By default, the extension is going to fetch resources on-demand from Keycloak where their URI are used to map the resources in your application that should be protected.

If you want to disable this behavior and fetch resources during startup, you can use the following configuration:

Note that, depending on how many resources you have in Keycloak the time taken to fetch them may impact your application startup time.

In the default configuration, Keycloak is responsible for managing the roles and deciding who can access which routes.

To configure the protected routes using the @RolesAllowed annotation or the application.properties file, check the Using OpenID Connect Adapter to Protect Jakarta REST Applications and Security Authorization guides. For more details, check the Security guide.

If you’d like to access a public resource without quarkus-keycloak-authorization trying to apply its policies to it then you need to create a permit HTTP Policy configuration in application.properties as documented in the Security Authorization guide.

Disabling a policy check using a Keycloak Authorization Policy such as:

is no longer required.

If you’d like to block access to the public resource to anonymous users then you can create an enforcing Keycloak Authorization Policy:

Note only the default tenant configuration applies when controlling anonymous access to the public resource is required.

In addition to resource permissions, you may want to specify method scopes. The scope usually represents an action that can be performed on a resource. You can create an enforcing Keycloak Authorization Policy with method scope like this:

Request path /api/protected/standard-way is now secured by the Keycloak Policy Enforcer and does not require any additions (such as @RolesAllowed annotation). In some cases, you may want to perform the same check programmatically. You are allowed to do that by injecting a SecurityIdentity instance in your beans, as demonstrated in the example below. Alternatively, if you annotate resource method with the @PermissionsAllowed annotation, you can achieve the same effect. The following example shows three resource method that all requires same 'read' scope:

It is possible to configure multiple policy enforcer configurations, one per each tenant, similarly to how it can be done for Multi-Tenant OpenID Connect Service Applications.

For example:

The configuration is based on the official Keycloak Policy Enforcer Configuration. If you are looking for more details about the different configuration options, please take a look at this documentation,