OpenID connect authentication

The OAuth2 OpenID Connect (OIDC) authentication is working in a way quite similar to Google (and GitHub) authentications, the only difference is that the authentication page cannot propose default values for the various endpoints, which have to be configured manually.

In case the web login will not be used, the client ID and client secret are not actually needed, and can be filled with two made up values (the validation just checks they are present, but they will be used only in the “authorisation flow”, but not when doing OGC requests where the client is supposed to have autonomously retrieved a valid bearer token).

The configuration GUI supports OpenID Discovery documents. If the server supports them it’s sufficient to provide the path to the document, or to the authentication service root, and the GUI will auto-fill itself based on the document contents:


The UI allows to set also the Post Logout Redirect URI which will be used to populate the post_logout_redirect_uri request param, when doing the global logout from the GeoServer UI. The OpenId provider will use the URI to redirect to the desired app page.

In addition, the OpenID connect authentication is able to extract the user roles from either the ID token or the Access Token:


The chosen attribute must be present in either the Access Token or in the Id token, and be either a string or an array of strings.

From the UI it is also possible to set the Response Mode value. The field can be kept empty, but it is needed when the OpenId server is used, as the Identity Provider doesn’t send the authorization code by default as a query string (that is mandatory in order to allow GeoServer and OpenId integration to work properly).

Finally, the admin can allow the sending of the client_secret during an access_token request through the Send Client Secret in Token Request. Some OpenId implementations requires it for the Authorization Code flow when the client app is a confidential client and can safely store the client_secret.

Logging OAuth2 Activity

The plugin includes an OIDC_LOGGING profile which is installed on startup. This logging profile quiets most GeoServer logging activity, while enabling trace logging for OAuth2 functionality.

The module also includes an additional connection setting to include the token details as additional log messages. This is intended to assist in troubleshooting during development and initial setup.


Log sensitive information

This setting can obviously be used to access sensitive information, and you are advised to clear logs after use.

To setup for troubleshooting OIDC activity:

  1. Navigate to Settings ‣ Global

  2. Select the logging profile OIDC_LOGGING

  3. Navigate Security ‣ Authentication

  4. Setup your OAuth2 OpenID Connect configuration with Log Sensitive Information (do not use in production) checked

  5. With these settings each individual step of the OAuth2 authentication is shown. The logging sensitive information setting logs access token and id token (the contents of these tokens may be decoded using

    DEBUG  [security.oauth2] - OIDC: - CLIENT_SECRET: squirrel
    DEBUG  [security.oauth2] - OIDC: received a CODE from Identity Provider - handing it in for ID/Access Token
    DEBUG  [security.oauth2] - OIDC: CODE=...
    DEBUG  [security.oauth2] - OIDC: Identity Provider returned Token, type=Bearer
    DEBUG  [security.oauth2] - OIDC: SCOPES=openid geocat
    DEBUG  [security.oauth2] - OIDC: ACCESS TOKEN: ....
    DEBUG  [security.oauth2] - OIDC: ID  TOKEN: ...
    DEBUG  [security.oauth2] - OIDC: Getting Roles from UserGroupService, location=null
    DEBUG  [security.oauth2] - OIDC: Geoserver Roles: ADMIN
    DEBUG  [security.oauth2] - OIDC: Geoserver Roles: ROLE_ADMINISTRATOR

OpenID Connect With Attached Access Bearer Tokens

The OpenID Connect plugin allows the use of Attached Bearer Access Tokens. This is typically used by automated (i.e. desktop or external Web Service) to access the Geoserver REST API.


Bearer Tokens

The setup process is as follows:

  1. Setup your OAuth2 OpenID Connect configuration as normal

  2. On the OpenID Connect configuration screen (bottom), makes sure “Allow Attached Bearer Tokens” is checked

  3. You cannot use ID Tokens as a Role Source for the attached Bearer Tokens (see below)

To Use:

  1. Obtain an Access Token from the underlying IDP

  2. Attach the access token to your HTTP request headers

Authorization: Bearer <token>

The Access Token (JWT) is validated;

  1. The Access Token is used to get the “userinfo” endpoint. The underlying IDP will verify the token (i.e. signature and expiry)

  2. The Audience of the Token is checked that it contains the GeoServer configured Client Id. This make sure an Access Token for another application is not being inappropriately reused in GeoServer (cf.

  3. The Subject of the userinfo and Access Token are verified to be about the same person. The OpenID specification recommends checking this (cf.

For KeyCloak, consider using the “userinfo endpoint” role source and configure Keycloak to put groups in the “userinfo.”

For Azure AD, configure Azure to allow access to the MS Graph API (memberOf) and use the “Microsoft Graph API (Azure AD)” role source.

To configure Azure AD for “memberOf” (“GroupMember.Read.All” permission) access;

  1. Go to your application in Azure AD (in the portal)

  2. On the left, go to “API permissions”

  3. Click “Add a permission”

  4. press “Microsoft Graph”

  5. press “Delegated permission”

  6. Scroll down to “GroupMember”

  7. Choose “GroupMember.Read.All”

  8. Press “Add permission”

  9. On the API Permission screen, press the “Grant admin consent for …” text

This has been tested with KeyCloak (with groups in the userinfo endpoint response), and with MS Azure AD (with the groups from the GraphAPI). This should work with other IDPs - however, make sure that the Subject and Audience token verification works with their tokens.

If you do not need Bearer Token functionality, it is recommended to turn this off.

Proof Key of Code Exchange (PKCE)

The OpenID Connect plugin allows the use of Proof Key of Code Exchange (PKCE).


Proof Key of Code Exchange

The setup process is as follows:

  1. Setup your OAuth2 OpenID Connect configuration as normal

  2. On the OpenID Connect configuration screen (bottom), makes sure “Use PKCE” is checked

To prevent client side request forgery:

  • Step 1: GeoServer will include a code_challenge during initial authorization code request

  • Step 2: GeoServer will include a code_verifer during the access token request.

    The authentication server will confirm that code_verifier hash matches the initial code_challenge in order the confirm the client is the same as in Step 1.

Log output of this exchange is as follows:

DEBUG  [oauth2.pkce] - Generate code_verifier: yQat4Y.....
DEBUG  [oauth2.pkce] - CODE_CHALLENGE: 5HiD...
DEBUG  [oauth2.pkce] - CODE_CHALLENGE_METHOD: S256
DEBUG  [oauth2.pkce] - CLIENT_SECRET: squirrel
DEBUG  [oauth2.pkce] - CODE_VERIFIER: yQat4Y...


JSON Web Key set URI

The JSON Web Key set URI provides the location of a document of public keys that can be used to check the signature of the provided accessToken.

Optional: It is no longer required to use Check Token Endpoint URL - if you leave that field blank you may rely only on the JSON Web Key set URI signature check. When use in this manner roles cannot be extracted from access token.

Enforce Token Validation

True by default.

Check this option to enforce the validation of the token signature.

Per the RFC 7517 or this doc from auth0, the parameters does not include neither public_key_use (but use nor key_id (but kid)

The RFC specifies that kid is optional (RFC 7517: JSON Web Key (JWK)) Use of this member is OPTIONAL.


Azure AD and ADFS setup

To make the OpenIdConnect filter to work properly with an Azure AD or ADFS server via the OpenId protocol, the user must set, in addition to the other configuration parameters, the Response Mode to query (otherwise by default ADFS will return a url fragment) and check the checkbox Send Client Secret in Token Request (the client_secret is mandatory in token request according to the Microsoft documentation).


SSL Trusted Certificates

When using a custom Keystore or trying to access a non-trusted or self-signed SSL-protected OAuth2 Provider from a non-SSH connection, you will need to add the certificates to the JVM Keystore.

In order to do this you can follow the next steps:

In this example we are going to

  1. Retrieve SSL certificates from Google domains:

    “Access Token URI” = therefore we need to trust or ( “Check Token Endpoint URL” = therefore we need to trust or (


    You will need to get and trust certificates from every different HTTPS URL used on OAuth2 Endpoints.

  2. Store SSL Certificates on local hard disk

  3. Add SSL Certificates to the Java Keystore

  4. Enable the JVM to check for SSL Certificates from the Keystore

  1. Retrieve the SSL Certificates from Google domains

    Use the openssl command in order to dump the certificate


    openssl s_client -connect

    And for

    openssl s_client -connect
  2. Store SSL Certificates on local hard disk

    Copy-and-paste the two sections -BEGIN CERTIFICATE-, -END CERTIFICATE- and save them into two different .cert files


    .cert file are plain text files containing the ASCII characters included on the -BEGIN CERTIFICATE-, -END CERTIFICATE- sections

    google.cert (or whatever name you want with .cert extension)


    google-apis.cert (or whatever name you want with .cert extension)

  3. Add SSL Certificates to the Java Keystore

    You can use the Java command keytool like this

    google.cert (or whatever name you want with .cert extension)

    keytool -import -noprompt -trustcacerts -alias google -file google.cert -keystore ${KEYSTOREFILE} -storepass ${KEYSTOREPASS}

    google-apis.cert (or whatever name you want with .cert extension)

    keytool -import -noprompt -trustcacerts -alias google-apis -file google-apis.cert -keystore ${KEYSTOREFILE} -storepass ${KEYSTOREPASS}

    or, alternatively, you can use some graphic tool which helps you managing the SSL Certificates and Keystores, like Portecle

    java -jar c:\apps\portecle-1.9\portecle.jar
  4. Enable the JVM to check for SSL Certificates from the Keystore

    In order to do this, you need to pass a JAVA_OPTION to your JVM:\tmp\keystore.key
  5. Restart your server


Here below you can find a bash script which simplifies the Keystore SSL Certificates importing. Use it at your convenience.

# get the SSL certificate
openssl s_client -connect ${HOST}:${PORT} </dev/null \
    | sed -ne '/-BEGIN CERTIFICATE-/,/-END CERTIFICATE-/p' > ${HOST}.cert

# create a keystore and import certificate
keytool -import -noprompt -trustcacerts \
    -alias ${HOST} -file ${HOST}.cert \
    -keystore ${KEYSTOREFILE} -storepass ${KEYSTOREPASS}

# verify we've got it.
keytool -list -v -keystore ${KEYSTOREFILE} -storepass ${KEYSTOREPASS} -alias ${HOST}