Service Providers

This page contains information about connecting services to EGI Check-in in order to allow user login through Check-in and to receive users’ attributes. Check-in is connected to a wide range of academic and social Identity Providers that users can choose from in order to access your service.

Services eligible for integration

EGI Operations, as owner of the Check-in service, must approve every request for integration of new services with Check-in. The approval (or non-approval) is based on some prerequisites, the relevance of the service for the EGI community and the available resources to support the integration. The prerequisites are described in the following sections.

EGI at any time can prevent a service provider to access the Check-in service

Services federated in EGI

All the services that are operated by Resource Providers federated in EGI federation and that abide to the RC OLA, and consequently to the relevant security policies of EGI, can be connected with Check-in. Fulfilling all the relevant EGI policies makes the service eligible in receiving attributes released by Check-in.

Services not federated in EGI

A service not part of the EGI federation can be integrated with Check-in if the organisation providing the service complies with the EGI security requirements relevant to the service providers.

By accepting the policies a service provider assures that they will operate the service in good faith, without deliberately exposing the user to security risks, without claiming intellectual property on the data owned by the user, and protecting sensitive data generated by the interaction of the user with the service.

The policies that the service provider will have to accept are available in the EGI Policies and procedures page and specifically are:

1. EGI Security Policy
2. Service Operations Security Policy
3. Traceability and Logging Policy
4. Security Incident Response Policy
5. Policy on the Processing of Personal Data

Service Provider integration workflow

To integrate your Service Provider with the EGI Check-in service, you need to create a registration request using the EGI Federation Registry Portal. You can also use the Federation Registry portal to request the reconfiguration or deregistration of an existing deployed service. Service registration requests typically require approval by an administrator. Please refer to the Federation Registry Documentation for more information.

The integration follows a two-step process:

1. Register your Service Provider and test integration with the demo instance of EGI Check-in by selecting the “Demo” integration environment during registration through the EGI Federation Registry Portal. Service registration requests require approval by an administrator. The review process for the demo environment involves primarily the technical aspects of the service configuration. However, moving the service to production requires compliance with all the eligibility criteria (see Step 2). The demo instance allows for testing authentication and authorisation through the academic and social Identity Providers connected to Check-in without affecting the production Check-in service. Note that while the demo instance has identical functionality to the production instance, no information is shared between the two systems.
   • You can also test new features of Check-in that are not available in production yet, by registering your Service Provider and testing integration with the development instance of Check-in. In the development instance service requests can be self-reviewed without the need to wait for approval from an administrator. As with the demo instance, the development instance allows for testing authentication and authorisation without affecting the production Check-in service. NB: the list of supported Identity Providers in the development instance is limited. Therefore, we recommend using any of the social identity providers or the EGI SSO to test the login workflow when using the development instance.
2. Register your Service Provider with the production instance of EGI Check-in by selecting the “Production” integration environment during registration through the EGI Federation Registry Portal. The production instance allows access to your service through the academic and social Identity Providers connected to Check-in. This requires that your service meets all the eligibility criteria and that integration has been thoroughly tested during Step 1.

General Information

EGI Check-in supports two authentication and authorisation protocols that you can choose from:

1. Security Assertion Markup Language (SAML) 2.0
2. OpenID Connect - an extension to OAuth 2.0

Service providers should ensure that a proper authorisation model is put in place: if low assurance accounts, like those coming from social media identity providers, are granted access without any vetting, it may lead to an abuse of their service.

Regardless of which of the two protocols you are going to use, you need to provide the following information to connect your service to EGI Check-in:

1. Name of the service (in English and optionally in other languages supported by the service)
2. Short description of the service
3. Site (URL) for localised information about the service; the content found at the URL SHOULD provide more complete information than what provided by the description
4. Contact information of the following types:
   • Helpdesk/Support contact information (for redirecting user)
   • Administrative
   • Technical
   • Security/incident response
5. Privacy statement URL: The privacy policy is used to document the data collected and processed by the service. You can use the Privacy Policy template
6. Logo URL (optional for showing in catalogues); if provided, logos SHOULD:
   • use a transparent background where appropriate to facilitate the usage of logos within a user interface
   • use PNG, or GIF (less preferred), images
   • use HTTPS URLs in order to avoid mixed-content warnings within browsers
   • have a size larger than 40000 and smaller than 50000 characters when encoded in base64
7. Country of the service
8. Compliance with the EGI Policies and the REFEDS Data Protection Code of Conduct

The most important URLs for each environment are listed in the table below but more information can be found in the protocol-specific sections that follow.

• Production
• Demo
• Development

SAML Service Provider

To enable federated access to a web-based application, you can connect to the EGI Check-in IdP as a SAML Service Provider (SP). Users of the application will be redirected to Check-in in order to log in, and Check-in can authenticate them using any of the supported backend authentication mechanisms, such as institutional IdPs registered with eduGAIN or Social Providers. Once the user is authenticated, EGI Check-in will return a SAML assertion to the application containing information about the authenticated user.

Metadata registration

SAML authentication relies on the use of metadata. Both parties (you as a SP and the EGI Check-in IdP) need to exchange metadata in order to know and trust each other. The metadata include information such as the location of the service endpoints that need to be invoked, as well as the certificates that will be used to sign SAML messages. The format of the exchanged metadata should be based on the XML-based SAML 2.0 specification. Usually, you will not need to manually create such an XML document, as this is automatically generated by all major SAML 2.0 SP software solutions (e.g., Shibboleth, SimpleSAMLphp, and mod_auth_mellon). It is important that you serve your metadata over HTTPS using a browser-friendly SSL certificate, i.e. issued by a trusted certificate authority.

You can get the metadata of the EGI Check-in IdP Proxy on a dedicated URL that depends on the integration environment being used:

• Production
• Demo
• Development

To register your SAML SP, you must submit a service registration request at Federation Registry. Your request should include the general information about your service (see General Information) and the SP’s metadata and entity ID.

Metadata

Metadata provided by your SP should contain a descriptive name of the service that your SP represents in at least English. It is recommended to also provide the name in other languages which are commonly used in the geographic scope of the deployment. The name should be placed in the <md:ServiceName> in the <md:AttributeConsumingService> container.

It is recommended that the <md:IDPSSODescriptor> element included in your SP metadata contains both an AuthnRequestsSigned and an WantAssertionsSigned attribute set to true.

Your SP metadata should also contain contact information for support and for a technical contact. The <md:EntityDescriptor> element should contain both a <md:ContactPerson> element with a contactType of "support" and a <md:ContactPerson> element with a contactType of "technical". The <md:ContactPerson> elements should contain at least one <md:EmailAddress>. The support address may be used for generic support questions about the service, while the technical contact may be contacted regarding technical interoperability problems. The technical contact must be responsible for the technical operation of the service represented by your SP.

Attributes

The EGI Check-in IdP is guaranteed to release a minimal subset of the REFEDS R&S attribute bundle to connected Service Providers without administrative involvement, subject to user consent. The following attributes constitute a minimal subset of the R&S attribute bundle:

• Community User Identifier (CUID) which is a globally unique, opaque, persistent and non-reassignable identifier identifying the user (voPersonID). For users whose community identity is managed by Check-in, this identifier is of the form <uniqueID>@egi.eu. The <uniqueID> portion is an opaque identifier issued by Check-in.
• Email address (mail)
• Display name (displayName) OR (givenName AND sn)

A more extensive list of all the attributes that may be made available to Service Providers is included in the User Attribute section.

Attribute-based authorisation

As mentioned in the General Information, omitting authorisation checks may lead to abuse of the service.

EGI Check-in provides information about the authenticated user that may be used by Service Providers in order to control user access to resources. This information is provided by the EGI Check-in IdP in the SAML attribute assertion. The table below lists the SAML attributes that are relevant for user authorisation:

References

• Shibboleth Service Provider Documentation
• SimpleSAMLphp Service Provider QuickStart
• Simple SAML 2.0 service provider with mod_auth_mellon Apache module

OpenID Connect Service Provider

Service Providers can be integrated with EGI Check-in using OpenID Connect (OIDC) as an alternative to the SAML2 protocol. To allow this, the EGI Check-in IdP provides an OpenID Connect (OAuth2) API based on Keycloak, which has been certified by the OpenID Foundation. Interconnection with the EGI Check-in OIDC Provider allows users to sign in using any of the supported backend authentication mechanisms, such as institutional IdPs registered with eduGAIN or Social Providers. Once the user has signed in, EGI Check-in can return OIDC Claims containing information about the authenticated user.

Client registration

Before your service can use the EGI Check-in OIDC Provider for user login, you must submit a service registration request using Federation Registry in order to obtain OAuth 2.0 credentials. The client configuration should include the general information about your service, as described in General Information section.

Obtaining OAuth 2.0 credentials

You need OAuth 2.0 credentials, which typically include a client ID and client secret, to authenticate users through the EGI Check-in OIDC Provider.

You can specify the client ID and secret when creating/editing your client or let them being automatically generated during registration (recommended).

To find the ID and secret of your client, do the following:

1. Select your client from the Manage Services Page.
2. Look for the Client ID in the Protocol tab.
3. Select the Display/edit client secret: option from the Protocol tab.

Setting one or more Redirection URIs

The Redirection URI(s) that you set when creating/editing your client determine where the EGI Check-in OIDC Provider sends responses to your authentication requests. Note that the Redirection URI MUST use the https scheme; the use of http Redirection URIs is only allowed in the development environment.

To find the Redirection URI(s) for your client, do the following:

1. Open the Manage Services
2. Find the redirect URIs for your client listed under the Protocol column of the overview table or Edit the particular client and look for the Redirect URI(s) in the Protocol tab.

Setting additional information about the client

It is strongly suggested that you add a short description and a logo for the client. Lastly, you need to set the email addresses of one or more contacts.

Claims

The EGI Check-in UserInfo Endpoint is an OAuth 2.0 Protected Resource that returns specific information about the authenticated end user as Claim Values. To obtain the requested Claims about the end user, the Client makes a request to the UserInfo Endpoint using an Access Token obtained through OpenID Connect Authentication. The scopes associated with the Access Token used to access the EGI Check-in UserInfo Endpoint will determine what Claims will be released. These Claims are represented by a JSON object that contains a collection of name and value pairs for the Claims.

The following scope values can be used to request Claims from the EGI Check-in UserInfo Endpoint:

A more extensive list of all the attributes that may be made available to Service Providers is included in the User Attribute section.

Grant Types

Check-in supports the following OpenID Connect/OAuth2 grant types:

• Authorization Code: used by Web Apps executing on a server.
• Token Exchange: used by clients to request and obtain security tokens in support of delegated access to resources.
• Device Code: used by devices that lack a browser to perform a user-agent based OAuth flow.

Endpoints

The most important OIDC/OAuth2 endpoints are listed below:

• Production
• Demo
• Development

Authorization Endpoint

The Authorization Endpoint performs Authentication of the end user. This is done by sending the User Agent to the Authorization Server's Authorization Endpoint for Authentication and Authorization, using request parameters defined by OAuth 2.0 and additional parameters and parameter values defined by OpenID Connect.

The request parameters of the Authorization endpoint are:

• client_id: ID of the client that ask for authentication to the Authorization Server.
• redirect_uri: URI to which the response will be sent.
• scope: A list of attributes that the application requires.
• state: Opaque value used to maintain state between the request and the callback.
• response_type: value that determines the authorization processing flow to be used. For Authorization Code grant set response_type=code. This way the response will include an authorization code.

Token Endpoint

To obtain an Access Token, an ID Token, and optionally a Refresh Token, the Client sends a Token Request to the Token Endpoint.

Depending on the grant type, the following parameters are required:

Authorization Code

Proof Key for Code Exchange (PKCE)

The Proof Key for Code Exchange (PKCE, pronounced pixie) extension (RFC 7636) describes a technique for public clients (clients without client_secret) to mitigate the threat of having the authorization code intercepted. The technique involves the client first creating a secret, and then using that secret again when exchanging the authorization code for an access token. This way if the code is intercepted, it will not be useful since the token request relies on the initial secret.

Client configuration

To enable PKCE you need to go to the Manage Services Page and create/edit a client. In “Protocol” tab under “Token Endpoint Authentication Method” select “No authentication” and in “Crypto” tab under “Proof Key for Code Exchange (PKCE) Code Challenge Method” select “SHA-256 hash algorithm”.

Protocol Flow

Because the PKCE-enhanced Authorization Code Flow builds upon the standard Authorization Code Flow, the steps are very similar.

First, the client creates and records a secret named the code_verifier. The code_verifier is a high-entropy cryptographic random STRING using the unreserved characters [A-Z] / [a-z] / [0-9] / “-” / “.” / “_” / “~”, with a minimum length of 43 characters and a maximum length of 128 characters. Then the client creates a code_challenge derived from the code_verifier by using one of the following transformations on the code verifier:

• plain code_challenge = code_verifier
• S256 code_challenge = BASE64URL-ENCODE(SHA256(ASCII(code_verifier)))

If the client is capable of using S256, it MUST use S256. Clients are permitted to use plain only if they cannot support S256 for some technical reason.

Then the code_challenge is sent in the Authorization Request along with the transformation method (code_challenge_method).

Example request:

GET "${AUTHORISATION_ENDPOINT}?
      client_id=${CLIENT_ID}
      &scope=openid%20profile%20email
      &redirect_uri=${REDIRECT_URI}
      &response_type=code
      &code_challenge=${CODE_CHALLENGE}
      &code_challenge_method=S256"

The Authorization Endpoint responds as usual but records code_challenge and the code_challenge_method.

Example response:

HTTP/1.1 302 Found
  Location: ${REDIRECT_URI}?
    code=fgtLHT

The client then sends the authorization code in the Access Token Request as usual but includes the code_verifier secret generated in the first request.

Example request:

curl -X POST "${TOKEN_ENDPOINT}" \
  -d "grant_type=authorization_code" \
  -d "code=${CODE}" \
  -d "client_id=${CLIENT_ID}" \
  -d "redirect_uri=${REDIRECT_URI}" \
  -d "code_verifier=${CODE_VERIFIER}" | python -m json.tool

The authorization server transforms code_verifier and compares it to code_challenge from the first request. Access is denied if they are not equal.

Example response:

{
  "access_token": "eyJraWQiOiJvaWRjIiwiYWxnIjoiUlMyNTYifQ...",
  "expires_in": 3599,
  "id_token": "eyJraWQiOiJvaWRjIiwiYWxnIjoiUlMyNTYifQ...",
  "scope": "openid email profile",
  "token_type": "Bearer"
}

Refresh request

The following request allows obtaining an access token from a refresh token using the grant_type value refresh_token:

Example request:

curl -X POST "${TOKEN_ENDPOINT}" \
  -u "${CLIENT_ID}":"${CLIENT_SECRET}" \
  -d "grant_type=refresh_token" \
  -d "refresh_token=${REFRESH_TOKEN}" \
  -d "scope=openid%20email%20profile" | python -m json.tool;

Example response:

{
  "access_token": "eyJraWQiOiJvaWRjIiwiYWx...",
  "expires_in": 3599,
  "id_token": "eyJraWQiOiJvaWRjIiwiYW...",
  "refresh_token": "eyJhbGciOiJub25...",
  "scope": "openid profile email",
  "token_type": "Bearer"
}

Refresh Request with PKCE

To combine the refresh token grant type with PKCE you need to make the following request:

curl -X POST "${TOKEN_ENDPOINT}" \
  -d "client_id=${CLIENT_ID}" \
  -d "grant_type=refresh_token" \
  -d "refresh_token=${REFRESH_TOKEN}" \
  -d "scope=openid%20email%20profile" | python -m json.tool;

Token Exchange

To get a token from client B using a token issued for client A, the parameters of the request are:

Example request:

curl -X POST "${TOKEN_ENDPOINT}" \
  -u "${CLIENT_B_ID}":"${CLIENT_B_SECRET}" \
  -d "grant_type=urn:ietf:params:oauth:grant-type:token-exchange" \
  -d "subject_token=${ACCESS_TOKEN_A}" \
  -d "subject_token_type=urn:ietf:params:oauth:token-type:access_token" \
  -d "scope=openid%20profile%20offline_access" | python -m json.tool;

Example response:

{
  "access_token": "eyJraWQiOiJvaWRjIiwiYWxnIjoiUl...",
  "expires_in": 3599,
  "id_token": "eyJraWQiOiJvaWRjIiwiYWxnIjoiUl...",
  "refresh_token": "eyJhbGciOiJub25lIn0.eyJleHAiO...",
  "scope": "openid profile offline_access",
  "token_type": "Bearer"
}

Device Code

The device code flow enables OAuth clients on (input-constrained) devices to obtain user authorization for accessing protected resources without using an on-device user-agent, provided that they have an internet connection.

1. Device Authorization Request

The client initiates the authorization flow by requesting a set of verification codes from the authorization server by making an HTTP “POST” request to the device authorization endpoint. The client constructs the request with the following parameters:

Example request:

curl -X POST "${DEVICE_CODE_ENDPOINT}" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "client_id=${CLIENT_ID}" \
  -d "client_secret=${CLIENT_SECRET}" \
  -d "scope=openid%20email%20profile" | python -m json.tool

Example response:

{
  "device_code": "HvtHOpSah_Anupq-0dtzvN7cb-wcnwxytiMzpBZBN6E",
  "expires_in": 600,
  "interval": 5,
  "user_code": "NMEM-SDPK",
  "verification_uri": "https://aai.egi.eu/auth/realms/egi/device",
  "verification_uri_complete": "https://aai.egi.eu/auth/realms/egi/device?user_code=NMEM-SDPK"
}

2. User Interaction

After receiving a successful Authorization Response, the client displays or otherwise communicates the user_code and the verification_uri to the end user and instructs them to visit the URI in a user agent on a secondary device (for example, in a browser on their mobile phone), and enter the user code.

3. Device Access Token Request

After displaying instructions to the user, the client makes an Access Token Request to the token endpoint. The request contains the following parameters:

Example request:

curl -X POST "${TOKEN_ENDPOINT}" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "grant_type=urn%3Aietf%3Aparams%3Aoauth%3Agrant-type%3Adevice_code" \
  -d "device_code=${DEVICE_CODE}" \
  -d "client_id=${CLIENT_ID}" \
  -d "client_secret=${CLIENT_SECRET}" \
  -d "scope=openid%20profile" | python -m json.tool

Example response:

{
  "access_token": "eyJraWQiOiJyc2ExIiwiYWxnIjoiUlMyNTYifQ.eyJzdWIiOiJhZG1pbiIs...",
  "expires_in": 3599,
  "id_token": "eyJraWQiOiJyc2ExIiwiYWxnIjoiUlMyNTYifQ.eyJzdWIiOiI5MDM0Mi...",
  "scope": "openid profile",
  "token_type": "Bearer"
}

Device Code with PKCE

To combine Device Code flow with PKCE you need to make the following requests:

1 - Device Authorization Request:

curl -X POST "${DEVICE_CODE_ENDPOINT}" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "client_id=${CLIENT_ID}" \
  -d "client_secret=${CLIENT_SECRET}" \
  -d "scope=openid%20email%20profile" \
  -d "code_challenge=${CODE_CHALLENGE}" \
  -d "code_challenge_method=S256" | python -m json.tool

2 - Device Access Token Request

curl -X POST "${TOKEN_ENDPOINT}" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "grant_type=urn%3Aietf%3Aparams%3Aoauth%3Agrant-type%3Adevice_code" \
  -d "device_code=${DEVICE_CODE}" \
  -d "client_id=${CLIENT_ID}" \
  -d "client_secret=${CLIENT_SECRET}" \
  -d "code_verifier=${CODE_VERIFIER}" | python -m json.tool

Claims-based authorisation

As mentioned in the General Information, omitting authorisation checks may lead to abuse of the service.

EGI Check-in provides information about the authenticated user that may be used by Service Providers in order to control user access to resources. This information is provided by the EGI Check-in OIDC Provider in the form of OIDC claims. The table below lists the claims that are relevant for user authorisation:

Example OIDC Client

In this guide we will demonstrate how to install and configure a Simple OIDC Client.

Install simple-oidc-client-php

This guide assumes the Apache HTTP server has been installed and the document root is /var/www/html

Move to the apache document root and download and extract simple-oidc-client-php.tar.gz.

Configure Client

Login to the EGI Federation Registry

Then create a new service or edit your existing service. In General tab fill all the required fields. For Integration Environment select Demo. In Protocol Specific tab select as Protocol the OIDC Service and then in the Redirect URI(s) insert your simple-oidc-client-php URL (e.g. http://localhost/simple-oidc-client-php/refreshtoken.php). This URL must link to refreshtoken.php which is located in simple-oidc-client-php directory. Next, in Scope select the scopes that your service needs. Then, submit the form and and self approve it. Finally you should get a pair of Client ID and Client Secret.

Configure simple-oidc-client-php

Now that you have everything you need, you can configure your login settings. Go to your terminal and open config.php with your favourite text editor.

Example:

vi simple-oidc-client-php/config.php

Let’s go quickly through the settings:

• title required, the title on the navigation bar
• img required, the source of the logo
• scope_info optional, a message that informs the user for the application requirements
• issuer required, the base URL of our IdentityServer instance. This will allow oidc-client to query the metadata endpoint so it can validate the tokens
• client_id required, the ID of the client we want to use when hitting the authorization endpoint
• client_secret optional, a value the offers better security to the message flow
• pkceCodeChallengeMethod optional, a string that defines the code challenge method for PKCE. Choose between plain or S256.
• redirect_url required, the redirect URL where the client and the browser agree to send and receive correspondingly the code
• scopesDefine required, defines the scopes the client supports
• refresh_token_note optional, info for the refresh token
• access_token_note optional, info for the access token
• manage_token_note optional, message the informs the user where can manage his tokens
• manageTokens optional, URL of the manage tokens service
• sessionName required, define the name of the cookie session
• sessionLifetime required, define the duration of the session. This must be equal to the validity time of the access token.

You must change the followings options based on your Service configuration you setup earlier:

• issuer
• client_id
• client_secret
• redirect_url
• scopesDefine
• sessionName (based on the installation path of the portal)

An example configuration follows:

<?php
// index.php interface configuration
$title = "Generate Tokens";
$img = "https://clickhelp.co/images/feeds/blog/2016.05/keys.jpg";
$scope_info = "This service requires the following permissions for your account:";

// Client configuration
$issuer = "https://aai-demo.egi.eu/auth/realms/egi";
$client_id = "CHANGE_ME";
$client_secret = "CHANGE_ME";  // comment if you are using PKCE
// $pkceCodeChallengeMethod = "S256";   // uncomment to use PKCE
$redirect_url = "http://localhost/simple-oidc-client-php/refreshtoken.php";
// add scopes as keys and a friendly message of the scope as value
$scopesDefine = array(
    'openid' => 'log in using your identity',
    'email' => 'read your email address',
    'profile' => 'read your basic profile info',
);
// refreshtoken.php interface configuration
$refresh_token_note = "NOTE: New refresh tokens expire in 12 months.";
$access_token_note = "NOTE: New access tokens expire in 1 hour.";
$manage_token_note = "You can manage your refresh tokens in the following link: ";
$manageTokens = $issuer . "manage/user/services";
$sessionName = "simple-oidc-client-php";
$sessionLifetime = 60*60;  // must be equal to access token validation time in seconds

Client Migration to Keycloak

The migration guide below applies to OIDC clients registered in the Development, Demo and Production environments of Check-in.

Development and Demo: Beginning June 24, 2022, clients using the legacy Check-in OIDC endpoints will no longer be supported.

Production: Beginning October 21, 2022, clients using the legacy Check-in OIDC endpoints will no longer be supported.

How to Migrate your Service to Keycloak

All the clients that were registered in MITREid Connect have been moved to Keycloak preserving all the options (Client ID, Client Secret, Redirect URIs etc.), so you do not need to re-register your Service.

Endpoints

The first thing you need to do is to update the OIDC endpoints according to the Endpoints table. If the Application/Library supports Dynamic Discovery, then you need to update on the issuer. Otherwise, you need to update all the Endpoints separately.

Size of the Tokens

The size of the Access/Refresh Tokens that are issued by Keycloak is larger of the respective Tokens created by MITREid Connect. For example, the size of an Access Token is around 1400 characters, depending on the information that are included in the payload of the JWT. So make sure that your OIDC implementation can handle larger Tokens.

Logout

The Redirect URI query parameter in the logout request has been changed from redirect to post_logout_redirect_uri and must be URL encoded. Also, the value of the post_logout_redirect_uri must be defined in the Valid Redirect URIs of the Service configuration in the EGI Federation Registry.

Token Introspection

The Token Introspection is available to all the clients that are using any authentication method (client_secret_basic, client_secret_post, client_secret_jwt or private_key_jwt) (Confidential Clients) to the Token Endpoint. Public Clients (clients that do not use any authentication method) will not be able to get a successful response from the Introspection Endpoint. Saying that, the “Introspection” option in the EGI Federation Registry will be removed.

PKCE

If you are not using PKCE (Proof Key for Code Exchange), please make sure to disable the “PKCE Code Challenge Method” in the Service configuration in EGI Federation Registry, otherwise you will get the following HTTP response during the authentication flow:

error=invalid_request&error_description=Missing parameter: code_challenge_method

Device Code

If you are using a confidential client with the Device Code grant, please make sure that the client_secret is present in the request to the Device Code Endpoint either as HTTP Basic or HTTP POST parameter (see Device Authorization Request).

Token Exchange

If you are using the Token Exchange grant, please make sure that the audience (Optional) defines the logical name of the service that the token will be used for; when specified, it must match the client ID of a client registered in Check-in otherwise an invalid_client error is returned ("description": "audience not found")

Client Credentials

If you are using the Client Credentials grant, there is a minor change in the responses from userinfo and introspection endpoints. The Client ID of the client is not released as the sub claim any more and has replaced with by the client_id claim. The sub contains the identifier of the client which is unique, non-reassignable and scoped @egi.eu.

Obtain Refresh Tokens

If you have obtained an Refresh Token from EGI Check-in Token Portal or oidc-agent issued by the MITREid Connect instance, you will need to replace them by creating new Refresh Tokens issued by Keycloak.

• If you have obtained Refresh Tokens using the EGI Check-in Token Portal, please check the following table:

  Environment	URL
  Production	https://aai.egi.eu/token
  Demo	https://aai-demo.egi.eu/token
  Development	https://aai-dev.egi.eu/token

• If you have obtained Refresh Tokens using the oidc-agent, please use the following command:

  oidc-gen --pub --issuer <ISSUER> --scope ...
  

  Note

  You can find the ISSUER in the Endpoints table.

Common issues

Error messages referring to missing code_challenge, code_challenge_method or code_verifier HTTP parameter

If you get error messages containing the PKCE HTTP parameters, probably the PKCE mode is enabled in your Service Configuration but the Application is not performing the PKCE mode.

To solve this, you need to follow the steps below:

1. Login to Federation Registry
2. Open your Service Configuration
3. Click on the “Protocol Specific” tab and scroll down to “Proof Key for Code Exchange (PKCE) Code Challenge Method” and select “PKCE will not be used for this service”
4. Click on “Submit” to apply the reconfiguration request

Error messages referring to invalid_code

If you try to perform the Authorization Code flow and you get an invalid_code error message, probably the Application sends the authorization request to the Authorization Endpoint of the Keycloak based EGI Check-in OP and then sends the code to the Token Endpoint of the MITREid Connect based EGI Check-in OP or vice versa.

To fix this you need to verify that you have updated all the OIDC Endpoints with the Keycloak ones. You can find all the OIDC Endpoints of Keycloak in the Endpoint table.

Error messages referring to the redirect_uri

If you try to perform the Authorization Code flow and you get an invalid_redirect_uri error, probably the redirect_uri in the Authorization Request mismatches with the Allowed Redirect URIs in the Service Configuration.

To solve this, you need to follow the steps below:

1. Login to Federation Registry
2. Open your Service Configuration
3. Click on the “Protocol Specific” tab and in the “Redirect URI(s)” edit the URI.
4. Click on “Submit” to apply the reconfiguration request

UserInfo invalid_token or 401 Unauthorized error response

If you are trying to make a request to the UserInfo Endpoint and the response contains the invalid_token error message, probably you are using an invalid Token or the UserInfo endpoint is wrong.

To solve this, please make sure the that:

1. You have obtained an Keycloak issued Access Token and you make a request to the Keycloak based UserInfo Endpoint
2. You have added the Access Token to the Authorization header of the request

502 Bad Gateway error after redirecting back to the Service

If you are using NGINX as a Reverse Proxy, and you are getting the following error message in the logs:

upstream sent too big header while reading response header from upstream

Then you need to increase the size of the buffer by adding the following options in the vhost configuration:

proxy_buffers 4 256k;
proxy_buffer_size 128k;
proxy_busy_buffers_size 256k;

Integrating Science Gateways with RCauth for obtaining (proxy) certificates

In order for Science Gateways (VO portals) to obtain RFC proxy certificates derived from personal end-entity certificates, an EGI Science Gateway can make use of the IGTF-approved IOTA-type RCauth.eu online CA. The actual integration goes via an intermediary service, called a Master Portal. EGI is running two Master Portal instances, one development, one production instance.

• Production
• Development

Registering a client at the Master Portal

In order to register a new client for your VO portal go to:

• EGI Development instance: https://masterportal-pilot.aai.egi.eu/mp-oa2-server/register
• EGI Production instance: https://aai.egi.eu/mp-oa2-server/register

In order to get the client approved, send an email to the administrator of the EGI Master Portal using checkin-support <AT> mailman.egi.eu.

Detailed information

For further and detailed instructions on the integration flow, see the generic RCAuth.eu MasterPortal VOPortal integration guide

SSH key authentication for proxy retrieval

The EGI MasterPortal also allows users to authenticate using SSH key pair, in order to retrieve proxy certificates from the MasterPortal. Users need to first upload the public key via a self-service portal, https://aai.egi.eu/sshkeys/. About once a week they need to follow a web-flow to ensure a long-lived proxy certificate is present in MasterPortal, e.g. by going to https://aai.egi.eu/vo-portal/. They can then obtain a proxy certificate by doing

ssh proxy@ssh.aai.egi.eu

and storing the output in /tmp/x509up_u$(id -u)

Generic information for users on how to do this can be found at Instructions for end users on how to use the SSH key authN for proxy retrieval. Alternatively VO portals could implement such functionality themselves by using the API described at the Master Portal sshkey endpoint description.

User attributes

This section defines the attributes that can be made available to services connected to Check-in.

1. Community User Identifier

2. Display Name

3. Given Name

4. Family Name

5. Username

6. Email Address

7. Verified email flag

8. Verified email list

9. Affiliation

10. Groups

11. Capabilities

12. GOCDB Roles

13. Assurance

14. CertEntitlement

15. SSH Public Key

16. ORCID iD

User authorisation

The following information about the authenticated user can be provided by EGI Check-in in order to control user access to resources:

1. VO/group membership and role information about the authenticated user
2. Capabilities
3. Identity Assurance
4. GOCDB roles

VO/group membership and role information

Background

VO/group membership and role information is encoded in entitlements (eduPersonEntitlement attribute values in SAML or eduperson_entitlement claim in OIDC). These entitlements are typically used to indicate access rights to protected resources. Entitlements are multi-valued, with each value formatted as a URN.

Syntax

An entitlement value expressing group membership and role information has the following syntax (components enclosed in square brackets are OPTIONAL):

urn:mace:egi.eu:group:<GROUP>[:<SUBGROUP>*][:role=<ROLE>]#<GROUP-AUTHORITY>

where:

• <GROUP> is the name of a VO, research collaboration or a top level arbitrary group. <GROUP> names are unique within the urn:mace:egi.eu:group namespace;
• zero or more <SUBGROUP> components represent the hierarchy of subgroups in the <GROUP>; specifying sub-groups is optional
• the optional <ROLE> component is scoped to the rightmost (sub)group; if no group information is specified, the role applies to the VO
• <GROUP-AUTHORITY> is a non-empty string that indicates the authoritative source for the entitlement value. For example, it can be the FQDN of the group management system that is responsible for the identified group membership information

Example:

urn:mace:egi.eu:group:fedcloud.egi.eu:role=vm_operator#aai.egi.eu

Capabilities

Background

The user’s capability information is encoded in entitlements (eduPersonEntitlement attribute values in SAML or eduperson_entitlement claim in OIDC). These entitlements are typically used to indicate access rights to protected resources. Entitlements are multi-valued, with each value formatted as a URN following the syntax defined in AARC-G027.

Syntax

An entitlement value expressing a capability has the following syntax (components enclosed in square brackets are OPTIONAL):

<NAMESPACE>:res:<RESOURCE>[:<CHILD-RESOURCE>]...[:act:<ACTION>[,<ACTION>]...]#<AUTHORITY>

where:

• <NAMESPACE> is controlled by the e-infrastructure, research infrastructure or research collaboration that manages the capability. The <NAMESPACE> of capabilities managed by Check-in is set to urn:mace:egi.eu, while, generally, it is in the form of urn:<NID>:<DELEGATED-NAMESPACE>[:<SUBNAMESPACE>]... where:

  • <NID> is the namespace identifier associated with a URN namespace registered with IANA2, ensuring global uniqueness. Implementers SHOULD use one of the existing registered URN namespaces, such as urn:mace[MACE].

  • <DELEGATED-NAMESPACE> is a URN sub-namespace delegated from one of the IANA registered NIDs to an organisation representing the e-infrastructure, research infrastructure or research collaboration. It is RECOMMENDED that a publicly accessible URN value registry for each delegated namespace be provided.

• The literal string "res" indicates that this is a resource-specific entitlement as opposed to, for example, an entitlement used for expressing group membership AARC-G002.

• <RESOURCE> is the name of the resource. Whether the name should be unique is an implementation decision.

• An optional list of colon-separated <CHILD-RESOURCE> components represents a specific branch of the hierarchy of resources under the identified <RESOURCE>.

• An optional list of comma-separated <ACTION>s MAY be included, which, if present, MUST be prefixed with the literal string “act”. This component MAY be used for further specifying the actions a user is entitled to do at a given resource. Note that the list of <ACTION>s is scoped to the rightmost child-resource; if no child-resource information is specified, actions apply to the top level resource. The interpretation of a capability without actions specified is an implementation detail.

• <AUTHORITY> is a mandatory and non-empty string that indicates the authoritative source of the capability. This SHOULD be used to further specify the exact issuing instance. For example, it MAY be the FQDN of the service that issued that specific capability. The <AUTHORITY> is specified in the f-component RFC8141 of the URN; thus, it is introduced by the number sign ("#") character and terminated by the end of the URN. All characters must be encoded according to RFC8141. Hence, the <AUTHORITY> MUST NOT be considered when determining equivalence (Section 3 in RFC8141) of URN-formatted capabilities. The <AUTHORITY> of capabilities managed by Check-in is typically set to aai.egi.eu.

Example:

urn:mace:egi.eu:res:rcauth#aai.egi.eu

Identity Assurance

Based on the authentication method selected by the user, the EGI proxy assigns a Identity Assurance, which is conveyed to the SP through both the eduPersonAssurance attribute and the Authentication Context Class (AuthnContextClassRef) of the SAML authentication response. EGI Check-in uses Assurance Profiles which distinguish between three Identity Assurance levels, similarly to the eID Assurance Framework (eIDAF). Each level is represented by a URI as follows:

• Low: Authentication through a social identity provider or other low identity assurance provider: https://aai.egi.eu/LoA#Low
• Substantial: Password/X.509 authentication at the user's home IdP: https://aai.egi.eu/LoA#Substantial
• High: Substantial + multi-factor authn (not yet supported, TBD): https://aai.egi.eu/LoA#High

Moreover, EGI Check-in follows the REFEDS Assurance framework (RAF). The EGI Check-in conveys any RAF values provided by the IdP directly to the SP, through the aforementioned methods. Furthermore, Check-in will append into the User’s profile any additional LoA, if the user is eligible for it. For example, a user having a Verified Email is eligible for the RAF value https://refeds.org/assurance/IAP/low

Some EGI SPs have been configured to provide limited access (or not to accept at all) credentials with the Low LoA.

Note: When logging in through the EGI SSO IdP, the LoA is determined based on the selected authentication mechanism as follows:

• Username/password credentials → Low
• X.509 certification → Substantial

GOCDB Roles

Background

GOCDB roles, as per GOCDB documentations, are encoded (eduPersonEntitlement attribute values in SAML or eduperson_entitlement claim in OIDC). These entitlements are typically used to indicate access rights to protected resources. Entitlements are multi-valued, with each value formatted as a URN.

Syntax

An entitlement value expressing GOCDB roles has the following syntax (components enclosed in square brackets are OPTIONAL):

urn:mace:egi.eu:goc.egi.eu:<PRIMARY_KEY>:<ON_ENTITY>:<USER_ROLE>@egi.eu

where:

• <PRIMARY_KEY> is the primary key for the user role, e.g. “123G0”
• <ON_ENTITY> is the name of the entity on which the user role applies to, e.g. “GRIDOPS-GOCDB”
• <USER_ROLE> is the user’s role, e.g. “Site Operations Manager”

Example:

urn:mace:egi.eu:goc.egi.eu:100453G0:GRIDOPS-CheckIn:Site+Administrator@egi.eu