- Docker installed and available on your machine
There are only four roles in OAuth2 and they are the following, please keep these four roles in mind as you read through the document:
-
Resource Owner: An entity capable of granting access to a protected resource. When the Resource Owner is a person, it is referred to as an end-user.
-
Resource Server: The server hosting the protected resources, capable of accepting and responding to protected resource requests using Access Tokens.
-
Client: An application making protected resource requests on behalf of the Resource Owner and with its authorization. The term "client" does not imply any particular implementation characteristics (e.g., whether the application executes on a server, a desktop, or other devices).
-
Authorization Server: The Authorization Server issues Access Tokens to the client.
We are going to go through each of these in the workshop to understand how each part works. Firstly how does OAuth2 work?
OAuth2 is an authorization framework, it allows a Resource Owner to authorize a client to access their resources. A real world example: When I use LinkedIn, it requests to access my GMail email contacts, this is OAuth2. LinkedIn is the client, I am the Resource Owner, gmail is the Resource Server and Google is the Authorization Server.
OAuth 2 has something called an abstract protocol. It is an outline of how the OAuth2 grants work. You can see this here:
Taken from https://www.rfc-editor.org/rfc/rfc6749#section-1.2
Let us look at the story of accessing GMail contacts from Linkedin.
- The Client (LinkedIn) asks the Resource Owner (User) if they can access their GMail contacts, this is the authorization request.
- The Resource Owner (User) either grants or denies access to the Client (LinkedIn)
- The Client (LinkedIn) presents to the Authoirzation Server (Google) that they have been granted access by the Resource Owner (User)
- The Client (LinkedIn) is issued an Access Token by the Authorization Server (Google)
- The Client (LinkedIn) uses the Access Token to request from the Resource Server (GMail) access the the Resource Ownerns (Users) GMail contacts
- The Resource Server (GMail) returns the GMail contacts to the Client (LinkedIn)
This is an outline of the steps for OAuth2 grants and how they work. As was mentioned, this is an abstract protocol, and the implementations of these protocols are called grants.
A key point is that with OAuth2 the different grants provide different ways to obtain an Access Token. What you may notice is that in the cases we look at, step E and F are constant throughout all the grants. The way to use an Access token is consistent.
The Authorization Code Grant (also known as Auth Code Grant) https://www.rfc-editor.org/rfc/rfc6749#section-4.1 is the most common grant. It follows the abstract protocol in terms of its mechanism and is used when human users are involved, these are the Resource Owners. Auth Code Grant is one of the most secure grants within OAuth2.
We can see in the following image how this works, note that the User-Agent is a browser, this whole grant is based on browser redirects:
This image is taken from https://www.rfc-editor.org/rfc/rfc6749#section-4.1
This grant starts with the user being directed to the authorization endpoint of the Authorization Server, this includes who the client is and where to send the authoirzation code. The Resource Owner authorizes (or denies) that the client can access its resources, an Authorization Code is then issued to the client. Once the client has the Authorization Code it can then send a request through a back channel (it is a server so the user will not see it), to the Authorization Servers token endpoint to request an Access Token. The request for the Access token consists of an Authorization Code, a redirect URL, a client id and client secret. Having an Authorization Code is not enough to prove who you are. It is also required for the client to authenticate itself to the Authorization Server. Once the Authorization Server has all this information and it is valid it can then issue an Access Token.
The Authorization Code Grant requires that the client is a confidential client. A confidential client is a client that can maintain a secret and will not leak it. This means a SPA can not use Auth Code Grant in this form, please check Auth Code Grant + PKCE to achieve this.
Client Credentials Grant is used for machine-to-machine communication. This grant has no resource owners involved, you can see from the following image:
This image is taken from https://www.rfc-editor.org/rfc/rfc6749#section-4.4
We see here that this image only shows how to obtain an Access Token. The client authenticates to the Authorization Server using a secret (like a password) to request an Access Token. As the client is requesting access on behalf of itself there is no Resource Owner needed. Once the client has the Access Token it can then send a request to the Resource Server.
Now we have seen how OAuth2 works and understand two of the grants (there are more), we should see how this works in practice. Let's start by creating our Authorization Server.
For our workshop we are going to use Keycloak. Keycloak is an opensource identity provider (IDP) https://www.keycloak.org/.
Lets fire it up.
docker run -d -p 8080:8080 -e KEYCLOAK_ADMIN=admin -e KEYCLOAK_ADMIN_PASSWORD=admin quay.io/keycloak/keycloak:19.0.3 start-dev
Give the image a minute or two to load and then visit the admin portal http://localhost:8080/
As we have seen, an Authorization Server issues Access Tokens to the client. The Authorization Server needs to keep track of the clients, i.e they need to be known to the Authorization Server. If the Authorization Server does not know who the clients are then how can it issue Access Tokens?
- First lets create a new realm, click the
master
dropdown can selectCreate Realm
- Enter the Realm name
oauth2-workshop
and selectCreate
We will use this realm (environment) to create our clients and manage the Authorization Server.
NOTE: In your own time explore of the realm and see what Keycloak offers.
Once the realm is created it creates an endpoint that describes what functionality exists from the provider.
You can see it at localhost:8080/realms/oauth2-workshop/.well-known/openid-configuration (if you called your realm something different replace the oauth2-workshop
).
You should see the following endpoints:
- authorization_endpoint
- token_endpoint
These are core to OAuth2, the authorization endpoint is used the obtain a grant from the Resource Owner. The token endpoint is used by the client to get an Access Token.
We have our Authorization Server setup. The first case we will look at is Client Credentials Grant. Lets create a Client
As we have seen for the Client Credentials Grant, a Client directly requests Access Tokens. We need to create the client.
- Select
Clients
from the sidebar and pressCreate client
- Enter the following information
- Client type:
OpenID Connect
- Client ID:
ClientCredentialsApp
- Name:
client credentials Application
- Description:
An OAuth2 client application using the client credentials grant
Success you have created a client application.
Next we need some credentials, select the Credentials
tab and reveal the Client secret
and note it down.
We now have everything for a Client Application.
We are going to tidy up a couple of things, so next select the Client scopes
tab and set everything the following to default:
Lets verify it is working correctly by requesting an Access Token from the token endpoint
CLIENT_CRED_ID=ClientCredentialsApp
CLIENT_CRED_SECRET=C2sBktd5E2yMjF7xu6ysNxYky3lRlOE6
curl --location -X POST 'http://localhost:8080/realms/oauth2-workshop/protocol/openid-connect/token' \
--header 'Content-Type: application/x-www-form-urlencoded' \
--data-urlencode "client_id=$CLIENT_CRED_ID" \
--data-urlencode "client_secret=$CLIENT_CRED_SECRET" \
--data-urlencode 'grant_type=client_credentials' \
| jq -r '.access_token'
You should get back an Access Token that will look something like:
eyJhbGciOiJSUzI1NiIsInR5cCIgOiAiSldUIiwia2lkIiA6ICJETmlGZUdIZ0VUdmNqdGJ4bzZVa0VIR3VfeXJrYlpCUnVJbzBGLWplWmtFIn0.eyJleHAiOjE2NjY4NzE5NzYsImlhdCI6MTY2Njg3MTY3NiwianRpIjoiY2FhNzdlNzAtNTc3Yi00NGRiLWJjYmEtNzI3M2Q1MzMyZTNiIiwiaXNzIjoiaHR0cDovL2xvY2FsaG9zdDo4MDgwL3JlYWxtcy9vYXV0aDItd29ya3Nob3AiLCJzdWIiOiJhYTY5OThjZC1iZTY4LTRkNTktODljYS00ZmJjMzlkM2E0NmYiLCJ0eXAiOiJCZWFyZXIiLCJhenAiOiJDbGllbnRDcmVkZW50aWFsc0FwcCIsImFjciI6IjEiLCJzY29wZSI6IiIsImNsaWVudEhvc3QiOiIxNzIuMTcuMC4xIiwiY2xpZW50SWQiOiJDbGllbnRDcmVkZW50aWFsc0FwcCIsImNsaWVudEFkZHJlc3MiOiIxNzIuMTcuMC4xIn0.CVprHTzQYneKvONa8HbjSLO1lp9S47mDlvMhTyfCJSKaFYBN84IqnbbjEkV4jLemkVmaXwDfSRFIzSG1JFO0tdH2DYPR-PjKg6SeLK1pp5MFdJra7TjgZe-Kq2PiDY_BcEHV0OyG7O9hnyZbsyujvI17xwBLSeAdU8qTNa85JrOdxfHx9kp_qpcvigLmTQr9BgCO_ckO_lb0ZI1uJOakaDWCewDspOFe6abazdaTgEraZOU_UROR7MYumSMq1xBJC79vuv3eMCzfjm3yKKXam05HuhbwdjCUI5_vIeL8DDtBtWV3FblKepj4K4i3fId_O6sd5XFyum6kVrXvKvFgAw
Success you have registered a client for client
We want to take the next step and create a client that uses the Auth Code Grant. To create a Client that supports this grant follow these steps.
- Select
Clients
from the sidebar and pressCreate client
- Enter the following information
- Client type:
OpenID Connect
- Client ID:
AuthCodeGrantApp
- Name:
Authorization Code Grant Application
- Description:
An OAuth2 client application using the client credentials grant
We have a Client and the Client needs to obtain an Access Token. We will setup a Client application that will direct the User (Resource Owner) to the keycloak authorization server to authenticate and request for access to their resources.
In this case the Client App wants to access the contacts of the Resource Owner from the Contacts API. To do this we will register a Scope into keycloak
- Visit the Client Scopes tab and select the
Create client scope
- Set the
Name
field asContacts-API
- Select
Save
- Select the
AuthCodeGrantApp
from the Client list - In the
Settings
tab scroll to theAccess settings
. Set theRoot URL
andHome URL
tohttp://localhost:7070
. TheValid redirect URIs
should be set tohttp://localhost:7070/login/oauth2/code/keycloak
, this is relative to your Root URL.
Note: The port value of 7070 is the port number you will be hosting your client application at. Please take a note if your change this.
- In the
Login Settings
set theLogin theme
to bebase
. Toggle on both theConsent required
andDisplay Client on screen
- Press
Save
to update the settings for the Client Application - We need to register that the Client can use the Contacts API, to do this we add it to the list of scopes available to it.
Visit the
Client Scopes
tab in the Client application. Select theAdd client scope
and in the dialoge box that appears select theContacts-API
and pressAdd
selecting theDefault
option.
The Client is now setup and ready to be used.
Final step before getting an Access Token is to create a User (Resource Owner).
- Select the
Users
and press theCreate new user
button. - Enter a value of your choice for the
Username
and pressCreate
- With the user created, navigate to the
Credentials
tab and pressSet password
. In the dialogue box enter a password of your choice. Make sure thatTemporary
is toggled to off
Next step is to get the credentials of the Client application. As we saw, for a Client to get an Access Token it is important that the Client can authenticate to the Authorization Server.
To get the credentials you can retrieve them from the Credentials
tab for the Client.
This can be seen in the following image
You have seen how to register clients in the Authorization Server. Next we need to serve up some resources, this if going to be a simple Spring API.
Navigate to <start.spring.io> choose your setup, however, you will need to select the following dependencies:
- Spring Web / Spring Reactive Web
- OAuth2 Resource Server
Download the project and open it up.
You should then create a controller
package and an Emails controller.
You can see an example of this setup in the sample Resource Server.
You will need to set a property for the Resource Server to act accordingly
spring.security.oauth2.resourceserver.jwt.jwk-set-uri=http://localhost:8080/realms/oauth2-workshop/protocol/openid-connect/certs
This tells Spring to retrieve the public keys from the Authorization Server, it can use these public keys to verify the JWT based Access Tokens. Spring does this automagically so you don't have to.
Note: You may need to change the server.port of the application, I've changed it to port 9090
In the controller the Principal is extracted and it is the JWT Access Token. It can be inspected to check claims in the token to extract information about the client.
@GetMapping
public List<EmailContact> getEmailContacts(JwtAuthenticationToken principal) {
if (!principal.getToken().getClaimAsString("scope").contains("Contacts-API")){
throw new AuthorizationServiceException("Not Authorized");
}
var clientEmails = clientEmailStore.get(principal.getToken().getClaimAsString("sub"));
if (clientEmails == null) {
throw new AuthorizationServiceException("Not Authorized");
}
return clientEmails;
}
Lets try this out using the client application from the Client Credentials, run the following request:
# Get the Access Token
CLIENT_CRED_ID=ClientCredentialsApp
CLIENT_CRED_SECRET=C2sBktd5E2yMjF7xu6ysNxYky3lRlOE6
ACCESS_TOKEN=$(curl --location -X POST 'http://localhost:8080/realms/oauth2-workshop/protocol/openid-connect/token' \
--header 'Content-Type: application/x-www-form-urlencoded' \
--data-urlencode "client_id=$CLIENT_CRED_ID" \
--data-urlencode "client_secret=$CLIENT_CRED_SECRET" \
--data-urlencode 'grant_type=client_credentials' \
| jq -r '.access_token')
# Call the Resource Server as the Client Application
curl -v --header "Authorization: Bearer $ACCESS_TOKEN" \
http://localhost:9090/emails
Success you should now see that your request returns a response.
Let's try this out using Auth Code Grant, simple visit http://localhost:7070/
Try it yourself:
- Try modifying your Access Token does the request still work?
- Wait till your Access Token expires, what happens when you send it to the Resource Server?
- Use a different JWKs and send your Access Token against the Resource Server. What happens?