Specification: Eclipse MicroProfile Interoperable JWT RBAC Version: 1.2-RC1 Status: Draft Release: July 21, 2020 Copyright (c) 2016-2017 Eclipse Microprofile Contributors: Red Hat Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Introduction
This specification outlines a proposal for using OpenID Connect(OIDC) based JSON Web Tokens(JWT) for role based access control(RBAC) of microservice endpoints.
Motivation
MicroProfile is a baseline platform definition that optimizes Enterprise Java for a microservices architecture and delivers application portability across multiple MicroProfile runtimes. While Java EE is a very feature rich platform and is like a toolbox that can be used to address a wide variety of application architectures, MicroProfile focuses on defining a small and a minimum set of Java EE standards that can be used to deliver applications based on a microservice architecture, they are:
-
JAX-RS
-
CDI
-
JSON-P
The security requirements that involve microservice architectures are strongly related with RESTful Security. In a RESTful architecture style, services are usually stateless and any security state associated with a client is sent to the target service on every request in order to allow services to re-create a security context for the caller and perform both authentication and authorization checks.
One of the main strategies to propagate the security state from clients to services or even from services to services involves the use of security tokens. In fact, the main security protocols in use today are based on security tokens such as OAuth2, OpenID Connect, SAML, WS-Trust, WS-Federation and others. While some of these standards are more related with identity federation, they share a common concept regarding security tokens and token based authentication.
For RESTful based microservices, security tokens offer a very lightweight and interoperable way to propagate identities across different services, where:
-
Services don’t need to store any state about clients or users
-
Services can verify the token validity if token follows a well known format. Otherwise, services may invoke a separated service.
-
Services can identify the caller by introspecting the token. If the token follows a well known format, services are capable to introspect the token by themselves, locally. Otherwise, services may invoke a separated service.
-
Services can enforce authorization policies based on any information within a security token
-
Support for both delegation and impersonation of identities
Today, the most common solutions involving RESTful and microservices security are based on OAuth2, OpenID Connect(OIDC) and JSON Web Tokens(JWT) standards.
Token Based Authentication
Token Based Authentication mechanisms allow systems to authenticate, authorize and verify identities based on a security token. Usually, the following entities are involved:
-
Issuer
-
Responsible for issuing security tokens as a result of successfully asserting an identity (authentication). Issuers are usually related with Identity Providers.
-
-
Client
-
Represented by an application to which the token was issued for. Clients are usually related with Service Providers. A client may also act as an intermediary between a subject and a target service (delegation).
-
-
Subject
-
The entity to which the information in a token refers to.
-
-
Resource Server
-
Represented by an application that is going to actually consume the token in order to check if a token gives access or not to a protected resource.
-
Independent of the token format or protocol in use, from a service perspective, token based authentication is based on the following steps:
-
Extract security token from the request
-
For RESTful services, this is usually achieved by obtaining the token from the Authorization header.
-
-
Perform validation checks against the token
-
This step usually depends on the token format and security protocol in use. The objective is make sure the token is valid and can be consumed by the application. It may involve signature, encryption and expiration checks.
-
-
Introspect the token and extract information about the subject
-
This step usually depends on the token format and security protocol in use. The objective is to obtain all the necessary information about the subject from the token.
-
-
Create a security context for the subject
-
Based on the information extracted from the token, the application creates a security context for the subject in order to use the information wherever necessary when serving protected resources.
-
Using JWT Bearer Tokens to Protect Services
For now, use cases are based on a scenario where services belong to the same security domain. This is an important note in order to avoid dealing with all complexities when you need to access services across different security domains. With that in mind, we assume that any information carried along with a token could be understood and processed (without any security breaches) by the different services involved.
The use case can be described as follows:
A client sends a HTTP request to Service A including the JWT as a bearer token:
GET /resource/1 HTTP/1.1 Host: example.com Authorization: Bearer mF_9.B5f-4.1JqM
On the server, a token-based authentication mechanism in front of Service A perform all steps described on the Token Based Authentication section. As part of the security context creation, the server establishes role and group mappings for the subject based on the JWT claims. The role to group mapping is fully configurable by the server along the lines of the Java EE RBAC security model.
JWT tokens follow a well defined and known standard that is becoming the most common token format to protect services. It not only provides a token format but additional security aspects like signature and encryption based on another set of standards like JSON Web Signature (JWS), JSON Web Encryption (JWE) and others.
There are few reasons why JWT is becoming so widely adopted:
-
Token validation doesn’t require an additional trip and can be validated locally by each service
-
Given its JSON nature, it is solely based on claims or attributes to carry authentication and authorization information about a subject.
-
Makes easier to support different types of access control mechanisms such as ABAC, RBAC, Context-Based Access Control, etc.
-
Message-level security using signature and encryption as defined by both JWS and JWE standards
-
Given its JSON nature, processing JWT tokens becomes trivial and lightweight. Especially if considering Java EE standards such as JSON-P or the different third-party libraries out there such as Nimbus, Jackson, etc.
-
Parties can easily agree on a specific set of claims in order to exchange both authentication and authorization information. Defining this along with the Java API and mapping to JAX-RS APIs are the primary tasks of the MP-JWT specification.
-
Widely adopted by different Single Sign-On solutions and well known standards such as OpenID Connect given its small overhead and ability to be used across different security domains (federation)
Recommendations for Interoperability
The maximum utility of the MicroProfile JWT(MP-JWT) as a token format depends on the agreement between both identity providers and service providers. This means identity providers - responsible for issuing tokens - should be able to issue tokens using the MP-JWT format in a way that service providers can understand in order to introspect the token and gather information about a subject. To that end, the requirements for the MicroProfile JWT are:
-
Be usable as an authentication token.
-
Be usable as an authorization token that contains Java EE application level roles indirectly granted via a groups claim.
-
Can be mapped to IdentityStore in JSR375.
-
Can support additional standard claims described in IANA JWT Assignments as well as non-standard claims.
To meet those requirements, we introduce 2 new claims to the MP-JWT:
-
"upn": A human readable claim that uniquely identifies the subject or user principal of the token, across the MicroProfile services the token will be accessed with.
-
"groups": A list of group names to be assigned to the principal of the MP-JWT. This typically will require a mapping at the application container level to application roles. A one-to-one mapping between group names and application role names is done by default when no custom mapping is defined in an implementation specific way.
Required and Recommended MP-JWT Headers and Claims
Required MP-JWT Headers
- alg
-
This JOSE header parameter identifies the cryptographic algorithm used to secure the JWT.
-
RSASSA-PKCS1-v1_5 SHA-256 or ECDSA using P-256 and SHA-256 algorithm are required when the claims have to be signed and must be specified as either "RS256", RFC7518, Section 3.3 or "ES256", RFC7515, Section 3.4.
-
RSAES using Optimal Asymmetric Encryption Padding algorithm is required when the claims or nested JWT tokens have to be encrypted. It is used for encrypting the content encryption key and must be specified as "RSA-OAEP", RFC7518, Section 4.3.
-
- enc
-
This JOSE header parameter is only required when the claims or nested JWT tokens have to be encrypted. It identifies the cryptographic algorithm used to encrypt the claims or nested JWT tokens. AES in Galois/Counter Mode (GCM) algorithm is required and must be specified as "A256GCM", RFC7518, Section 5.3
Recommended MP-JWT headers
- typ
-
This JOSE header parameter identifies the token as an RFC7519 and should be "JWT". RFC7519, Section 5.1
- kid
-
This JOSE header parameter is a hint indicating which key was used to secure the JWT. RFC7515, Section-4.1.4. It may need to be set if the verification key is in the JWK format.
Required MP-JWT claims
- iss
-
Identifies the token issuer
- iat
-
Identifies the time at which the JWT was issued. This claim can be used to determine the age of the JWT. Its value MUST be a number containing a NumericDate value. RFC7519, Section 4.1.6
- exp
-
Identifies the expiration time on or after which the JWT MUST NOT be accepted for processing. The processing of the "exp" claim requires that the current date/time MUST be before the expiration date/time listed in the "exp" claim. Implementers MAY provide for some small leeway, usually no more than a few minutes, to account for clock skew. Its value MUST be a number containing a NumericDate value. RFC7519, Section 4.1.4
- upn (or preferred_username or sub)
-
MP-JWT "upn" claim is the user principal name in the java.security.Principal interface, and is the caller principal name in javax.security.enterprise.identitystore.IdentityStore. If this claim is missing, fallback to the "preferred_username", OIDC Section 5.1 should be attempted, and if that claim is missing, fallback to the "sub" claim should be used.
Recommended MP-JWT claims
- sub
-
Uniquely identifies the principal that is the subject of the JWT. See the "upn" claim for how this relates to the container java.security.Principal. RFC7519, Section 4.1.2.
- jti
-
Provides a unique identifier for the JWT. The identifier value MUST be assigned in a manner that ensures that there is a negligible probability that the same value will be accidentally assigned to a different data object; if the application uses multiple issuers, collisions MUST be prevented among values produced by different issuers as well. The "jti" claim can be used to prevent the JWT from being replayed. The "jti" value is a case-sensitive string. RFC7519, Section 4.1.7
- aud
-
Identifies the MP JWT endpoint(s) which can be accessed by JWT. RFC7519, Section 4.1.3
[NOTE] MP JWT implementations may enforce that JWT tokens contain all the recommended headers and claims. The recommended headers and claims may become required in the future versions of the MP JWT specification.
[NOTE] It is recommended that JWT tokens have a groups
claim if the endpoint requires Authorization but MP JWT implementations can map the groups from the other claims if the tokens have been issued by OpenId Connect and other providers which currently do not support MP JWT.
If no groups information can be extracted directly from the groups
claim or via the custom mappers from other custom claims in a given token then this token can be accepted if the endpoint requires Authentication only.
Note
|
NumericDate used by exp , iat , and other date related claims is a JSON numeric value
representing the number of seconds from 1970-01-01T00:00:00Z UTC until the specified
UTC date/time, ignoring leap seconds
|
An example minimal MP-JWT in JSON would be:
{
"typ": "JWT",
"alg": "RS256",
"kid": "abc-1234567890"
}
{
"iss": "https://server.example.com",
"jti": "a-123",
"exp": 1311281970,
"iat": 1311280970,
"sub": "24400320",
"upn": "jdoe@server.example.com",
"groups": ["red-group", "green-group", "admin-group", "admin"],
}
This specification defines a JsonWebToken java.security.Principal interface extension that makes this set of required claims available via get style accessors. The JsonWebToken interface definition is:
package org.eclipse.microprofile.jwt;
public interface JsonWebToken extends Principal {
/**
* Returns the unique name of this principal. This either comes from the upn
* claim, or if that is missing, the preferred_username claim. Note that for
* guaranteed interoperability a upn claim should be used.
*
* @return the unique name of this principal.
*/
@Override
String getName();
/**
* Get the raw bearer token string originally passed in the authentication
* header
* @return raw bear token string
*/
default String getRawToken() {
return getClaim(Claims.raw_token.name());
}
/**
* The iss(Issuer) claim identifies the principal that issued the JWT
* @return the iss claim.
*/
default String getIssuer() {
return getClaim(Claims.iss.name());
}
/**
* The aud(Audience) claim identifies the recipients that the JWT is
* intended for.
* @return the aud claim.
*/
default Set<String> getAudience() {
return getClaim(Claims.aud.name());
}
/**
* The sub(Subject) claim identifies the principal that is the subject of
* the JWT. This is the token issuing
* IDP subject, not the
*
* @return the sub claim.
*/
default String getSubject() {
return getClaim(Claims.sub.name());
}
/**
* The jti(JWT ID) claim provides a unique identifier for the JWT.
The identifier value MUST be assigned in a manner that ensures that
there is a negligible probability that the same value will be
accidentally assigned to a different data object; if the application
uses multiple issuers, collisions MUST be prevented among values
produced by different issuers as well. The "jti" claim can be used
to prevent the JWT from being replayed.
* @return the jti claim.
*/
default String getTokenID() {
return getClaim(Claims.jti.name());
}
/**
* The exp (Expiration time) claim identifies the expiration time on or
* after which the JWT MUST NOT be accepted
* for processing in seconds since 1970-01-01T00:00:00Z UTC
* @return the exp claim.
*/
default long getExpirationTime() {
return getClaim(Claims.exp.name());
}
/**
* The iat(Issued at time) claim identifies the time at which the JWT was
* issued in seconds since 1970-01-01T00:00:00Z UTC
* @return the iat claim
*/
default long getIssuedAtTime() {
return getClaim(Claims.iat.name());
}
/**
* The groups claim provides the group names the JWT principal has been
* granted.
*
* This is a MicroProfile specific claim.
* @return a possibly empty set of group names.
*/
default Set<String> getGroups() {
return getClaim(Claims.groups.name());
}
/**
* Access the names of all claims are associated with this token.
* @return non-standard claim names in the token
*/
Set<String> getClaimNames();
/**
* Verify is a given claim exists
* @param claimName - the name of the claim
* @return true if the JsonWebToken contains the claim, false otherwise
*/
default boolean containsClaim(String claimName) {
return claim(claimName).isPresent();
}
/**
* Access the value of the indicated claim.
* @param claimName - the name of the claim
* @return the value of the indicated claim if it exists, null otherwise.
*/
<T> T getClaim(String claimName);
/**
* A utility method to access a claim value in an {@linkplain Optional}
* wrapper
* @param claimName - the name of the claim
* @param <T> - the type of the claim value to return
* @return an Optional wrapper of the claim value
*/
default <T> Optional<T> claim(String claimName) {
return Optional.ofNullable(getClaim(claimName));
}
/**
* A utility method to access a claim value in an {@linkplain Optional}
* wrapper
* @param claim - the claim
* @param <T> - the type of the claim value to return
* @return an Optional wrapper of the claim value
*/
default <T> Optional<T> claim(Claims claim) {
return claim(claim.name());
}
}
Additional Claims
The JWT can contain any number of other custom and standard claims, and these are made available from the JsonWebToken getOtherClaim(String) method. An example MP-JWT that contains additional "auth_time", "preferred_username", "acr", "nbf", "aud" and "roles" claims is:
{
"typ": "JWT",
"alg": "RS256",
"kid": "abc-1234567890"
}
{
"iss": "https://server.example.com",
"aud": ["s6BhdRkqt3"],
"exp": 1311281970,
"iat": 1311280970,
"sub": "24400320",
"upn": "jdoe@server.example.com",
"groups: ["red-group", "green-group", "admin-group"],
"roles": ["auditor", "administrator"],
"jti": "a-123",
"auth_time": 1311280969,
"preferred_username": "jdoe",
"acr": "phr",
"nbf": 1311288970
}
The Claims
Enumeration Utility Class, and the Set of Claim Value Types
The org.eclipse.microprofile.jwt.Claims
utility class encapsulate an enumeration
of all the standard JWT related claims along with a description and the required
Java type for the claim as returned from the JsonWebToken#getClaim(String)
method.
public enum Claims {
// The base set of required claims that MUST have non-null values in the JsonWebToken
iss("Issuer", String.class),
sub("Subject", String.class),
exp("Expiration Time", Long.class),
iat("Issued At Time", Long.class),
jti("JWT ID", String.class),
upn("MP-JWT specific unique principal name", String.class),
groups("MP-JWT specific groups permission grant", Set.class),
raw_token("MP-JWT specific original bearer token", String.class),
// The IANA registered, but MP-JWT optional claims
aud("Audience", Set.class),
nbf("Not Before", Long.class),
auth_time("Time when the authentication occurred", Long.class),
updated_at("Time the information was last updated", Long.class),
azp("Authorized party - the party to which the ID Token was issued", String.class),
nonce("Value used to associate a Client session with an ID Token", String.class),
at_hash("Access Token hash value", Long.class),
c_hash("Code hash value", Long.class),
full_name("Full name", String.class),
family_name("Surname(s) or last name(s)", String.class),
middle_name("Middle name(s)", String.class),
nickname("Casual name", String.class),
given_name("Given name(s) or first name(s)", String.class),
preferred_username("Shorthand name by which the End-User wishes to be referred to", String.class),
email("Preferred e-mail address", String.class),
email_verified("True if the e-mail address has been verified; otherwise false", Boolean.class),
gender("Gender", String.class),
birthdate("Birthday", String.class),
zoneinfo("Time zone", String.class),
locale("Locale", String.class),
phone_number("Preferred telephone number", String.class),
phone_number_verified("True if the phone number has been verified; otherwise false", Boolean.class),
address("Preferred postal address", JsonObject.class),
acr("Authentication Context Class Reference", String.class),
amr("Authentication Methods References", String.class),
sub_jwk("Public key used to check the signature of an ID Token", JsonObject.class),
cnf("Confirmation", String.class),
sip_from_tag("SIP From tag header field parameter value", String.class),
sip_date("SIP Date header field value", String.class),
sip_callid("SIP Call-Id header field value", String.class),
sip_cseq_num("SIP CSeq numeric header field parameter value", String.class),
sip_via_branch("SIP Via branch header field parameter value", String.class),
orig("Originating Identity String", String.class),
dest("Destination Identity String", String.class),
mky("Media Key Fingerprint String", String.class),
jwk("JSON Web Key Representing Public Key", JsonObject.class),
jwe("Encrypted JSON Web Key", String.class),
kid("Key identifier", String.class),
jku("JWK Set URL", String.class),
UNKNOWN("A catch all for any unknown claim", Object.class)
;
...
/**
* @return A desccription for the claim
*/
public String getDescription() {
return description;
}
/**
* The required type of the claim
* @return type of the claim
*/
public Class<?> getType() {
return type;
}
}
Note that the groups
and aud
claims should only be injected into a Set of String, not as a comma delimited string.
The current complete set of valid claim types is therefore, (excluding the invalid Claims.UNKNOWN Void type):
-
java.lang.String
-
java.lang.Long and long
-
java.lang.Boolean and boolean
-
java.util.Set<java.lang.String>
-
javax.json.JsonValue.TRUE/FALSE
-
javax.json.JsonString
-
javax.json.JsonNumber
-
javax.json.JsonArray
-
javax.json.JsonObject
Custom claims not handled by the Claims enum must be any of the types defined in the previous acceptable claim types list.
Service Specific Authorization Claims
An extended form of authorization on a per service basis using a "resource_access" claim has been postponed to a future release. See Future Directions for more information.
Marking a JAX-RS Application as Requiring MP-JWT Access Control
Since the MicroProfile does not specify a deployment format, and currently does
not rely on servlet metadata descriptors, we have added an org.eclipse.microprofile.jwt.LoginConfig
annotation that provides the same information as the web.xml login-config
element. It’s intended usage is to mark a JAX-RS Application
as requiring
MicroProfile JWT RBAC as shown in the following sample:
import org.eclipse.microprofile.annotation.LoginConfig;
import javax.ws.rs.ApplicationPath;
import javax.ws.rs.core.Application;
@LoginConfig(authMethod = "MP-JWT", realmName = "TCK-MP-JWT")
@ApplicationPath("/")
public class TCKApplication extends Application {
}
The MicroProfile JWT implementation is responsible for either directly processing this annotation, or mapping it to an equivalent form of metadata for the underlying implementation container.
Requirements for Rejecting MP-JWT Tokens
The MP-JWT specification requires that an MP-JWT implementation reject a JWT token as an invalid MP-JWT token if any of the following conditions are not met:
-
The JWT must have a JOSE "alg" header that indicates the token was signed using the RS256 or ES256 algorithm when the service endpoint expects signed tokens.
-
The JWT must have the JOSE "alg" and "enc" headers that indicate that the token was encrypted using the RSA-OAEP and A256GCM algorithms when the service endpoint expects encrypted tokens. If the encrypted content is a nested signed JWT token then it must meet the condition 1.
-
The JWT must have an "iss" claim representing the token issuer that maps to an MP-JWT implementation container runtime configured value. Any issuer other than those issuers that have been whitelisted by the container configuration must be rejected with an HTTP_UNAUTHENTICATED(401) error.
-
The JWT must have an "iat" claim representing the token issuance time. It is impossible to calculate how long the token may have been used for without the "iat" claim so such tokens must be rejected with an HTTP_UNAUTHENTICATED(401) error.
-
The JWT must have an "exp" claim representing the token expiration time. Tokens without the "exp" claim have an unlimited lifetime and must be rejected with an HTTP_UNAUTHENTICATED(401) error.
-
The token must have at least one of "upn" or "preferred_username" or "sub" claim for Java Principal to have a name. Tokens from which no Principal name can be deduced must be rejected with an HTTP_UNAUTHENTICATED(401) error.
-
The signed JWT token verification and encrypted JWT token decryption failures must lead to an HTTP_UNAUTHENTICATED(401) error.
Mapping MP-JWT Tokens to Java EE Container APIs
The requirements of how a JWT should be exposed via the various Java EE container APIs is discussed in this section. For the 1.0 release, the only mandatory container integration is with the JAX-RS container, and injection of the MP-JWT types.
CDI Injection Requirements
This section describes the requirements for MP-JWT implementations with regard to the injection of MP-JWT tokens and their associated claim values.
Injection of JsonWebToken
An MP-JWT implementation must support the injection of the currently authenticated
caller as a JsonWebToken
with @RequestScoped
scoping which must work even if the outer bean is @ApplicationScoped
:
@Path("/endp")
@DenyAll
@ApplicationScoped
public class RolesEndpoint {
@Inject
private JsonWebToken callerPrincipal;
If there is no JWT in the request, an empty JsonWebToken
is injected, which means all method calls to this token
return null
. Note that MP JWT will still perform Authentication and Authorization if the endpoint requires these
verifications. Effectively, the injected empty token is only visible on unauthenticated and unauthorized endpoints.
If a JWT is sent to an endpoint that does not require Authentication and/or Authorization then it still must be
verified before it can be accessed via JsonWebToken
interface.
Endpoints which need to control the authentication process themselves can check if a token is available by calling
a JsonWebToken.getRawToken()
method.
Injection of JsonWebToken
claims via Raw Type, ClaimValue
, javax.enterprise.inject.Instance
and JSON-P Types
This specification requires support for injection of claims from the current
JsonWebToken
using the org.eclipse.microprofile.jwt.Claim
qualifier:
/**
* Annotation used to signify an injection point for a {@link ClaimValue} from
* a {@link JsonWebToken}
*/
@Qualifier
@Retention(RetentionPolicy.RUNTIME)
@Target({ElementType.FIELD, ElementType.METHOD, ElementType.PARAMETER, ElementType.TYPE})
public @interface Claim {
/**
* The value specifies the id name the claim to inject
* @return the claim name
* @see JsonWebToken#getClaim(String)
*/
@Nonbinding
String value() default "";
/**
* An alternate way of specifying a claim name using the {@linkplain Claims}
* enum
* @return the claim enum
*/
@Nonbinding
Claims standard() default Claims.UNKNOWN;
}
with @Dependent
scoping.
MP-JWT implementations are required to throw a DeploymentException
when detecting the ambiguous use of a
@Claim
qualifier that includes inconsistent non-default values for both the value and standard elements as
is the case shown here:
@ApplicationScoped
public class MyEndpoint {
@Inject
@Claim(value="exp", standard=Claims.iat)
private Long timeClaim;
...
}
The current complete set of valid claim types is:
-
java.lang.String
-
java.lang.Long and long
-
java.lang.Boolean and boolean
-
java.util.Set<java.lang.String>
-
javax.json.JsonValue.TRUE/FALSE
-
javax.json.JsonString
-
javax.json.JsonNumber
-
javax.json.JsonArray
-
javax.json.JsonObject
-
java.util.Optional wrapper of the above types.
-
org.eclipse.microprofile.jwt.ClaimValue wrapper of the above types.
MP-JWT implementations are required to support injection of the claim values using any of these types. The claims are automatically converted to the type used in the injection point where the type must be any of the types defined in the previous acceptable claim types list.
The org.eclipse.microprofile.jwt.ClaimValue
interface is:
/**
* A representation of a claim in a {@link JsonWebToken}
* @param <T> the expected type of the claim
*/
public interface ClaimValue<T> extends Principal {
/**
* Access the name of the claim.
* @return The name of the claim as seen in the JsonWebToken content
*/
@Override
public String getName();
/**
* Access the value of the claim.
* @return the value of the claim.
*/
public T getValue();
}
The following example code fragment illustrates various
examples of injecting different types of claims using a range of generic forms of
the ClaimValue
, JsonValue
as well as the raw claim types:
import org.eclipse.microprofile.jwt.Claim;
import org.eclipse.microprofile.jwt.ClaimValue;
import org.eclipse.microprofile.jwt.Claims;
@Path("/endp")
@DenyAll
@RequestScoped
public class RolesEndpoint {
...
// Raw types
@Inject
@Claim(standard = Claims.raw_token)
private String rawToken;
@Inject (1)
@Claim(standard=Claims.iat)
private Long issuedAt;
// ClaimValue wrappers
@Inject (2)
@Claim(standard = Claims.raw_token)
private ClaimValue<String> rawTokenCV;
@Inject
@Claim(standard = Claims.iss)
private ClaimValue<String> issuer;
@Inject
@Claim(standard = Claims.jti)
private ClaimValue<String> jti;
@Inject (3)
@Claim("jti")
private ClaimValue<Optional<String>> optJTI;
@Inject
@Claim("jti")
private ClaimValue objJTI;
@Inject
@Claim("groups")
private ClaimValue<Set<String>> groups;
@Inject (4)
@Claim(standard=Claims.iat)
private ClaimValue<Long> issuedAtCV;
@Inject
@Claim("iat")
private ClaimValue<Long> dupIssuedAt;
@Inject
@Claim("sub")
private ClaimValue<Optional<String>> optSubject;
@Inject
@Claim("auth_time")
private ClaimValue<Optional<Long>> authTime;
@Inject (5)
@Claim("custom-missing")
private ClaimValue<Optional<Long>> custom;
//
@Inject
@Claim(standard = Claims.jti)
private Instance<String> providerJTI;
@Inject (6)
@Claim(standard = Claims.iat)
private Instance<Long> providerIAT;
@Inject
@Claim("groups")
private Instance<Set<String>> providerGroups;
//
@Inject
@Claim(standard = Claims.jti)
private JsonString jsonJTI;
@Inject
@Claim(standard = Claims.iat)
private JsonNumber jsonIAT;
@Inject (7)
@Claim("roles")
private JsonArray jsonRoles;
@Inject
@Claim("customObject")
private JsonObject jsonCustomObject;
-
Injection of a non-proxyable raw type like java.lang.Long must happen in a RequestScoped bean as the producer will have dependendent scope.
-
Injection of the raw MP-JWT token string.
-
Injection of the jti token id as an
Optional<String>
wapper. -
Injection of the issued at time claim using an
@Claim
that references the claim name using the Claims.iat enum value. -
Injection of a custom claim that does exist will result in an Optional<Long> value for which isPresent() will return false.
-
Another injection of a non-proxyable raw type like java.lang.Long, but the use of the javax.enterprise.inject.Instance interface allows for injection to occur in non-RequestScoped contexts.
-
Injection of a JsonArray of role names via a custom "roles" claim.
The example shows that one may specify the name of the claim using a
string or a Claims
enum value. The string form would allow for specifying non-standard
claims while the Claims
enum approach guards against typos.
Handling of Non-RequestScoped Injection of Claim Values
MP-JWT implementations are required to support a claim value injection into @ApplicationScoped scoped beans. A warning may be issued when the injection point is not an org.eclipse.microprofile.jwt.ClaimValue
or javax.inject.Provider
interface.
MP-JWT implementations are required to generate a javax.enterprise.inject.spi.DeploymentException
for a claim value injection into Passivation capable beans, for example, @SessionScoped.
MP JWT implementations may issue a warning for any other context with a lifetime greater than @RequestScoped.
Note
|
If one needs to inject a claim value into a scope with a lifetime greater than @RequestScoped, such as @ApplicationScoped or @SessionScoped, one can also use the javax.enterprise.inject.Instance interface to do so. |
JAX-RS Container API Integration
The behavior of the following JAX-RS security related methods is required for MP-JWT implementations.
Using the Common Security Annotations for the Java Platform (JSR-250)
The expectations for use of the various security annotations described in sections 2.9 - 2.12 of JSR-250 (@RolesAllowed, @PermitAll, @DenyAll), is that MP-JWT containers support the behavior as described in those sections. In particular, the interaction between the annotations should be as described in section 2.12 of JSR-250.
Mapping the @RolesAllowed
to the MP-JWT group claim
In terms of mapping between the MP-JWT claims and role names used in @RolesAllowed, the role names that have been mapped to group names in the MP-JWT "groups" claim, MUST result in an allowing authorization decision wherever the security constraint has been applied.
Recommendations for Optional Container Integration
This section describes the expected behaviors for Java EE container APIs other than JAX-RS.
javax.security.enterprise.identitystore.IdentityStore.getCallerGroups(CredentialValidationResult)
This method should return the set of names found in the "groups" claim in the JWT if it exists, an empty set otherwise.
javax.ejb.SessionContext.getCallerPrincipal()
The java.security.Principal returned from this method MUST be an instance of org.eclipse.microprofile.jwt.JsonWebToken
.
javax.ejb.SessionContext#isCallerInRole(String)
This method MUST return true for any name that is included in the MP-JWT "groups" claim, as well as for any role name that has been mapped to a group name in the MP-JWT "groups" claim.
Overriding @LoginConfig from web.xml login-config
If a deployment with a web.xml descriptor contains a login-config element, an MP-JWT implementation should view the web.xml metadata as an override to the deployment annotation.
javax.servlet.http.HttpServletRequest.getUserPrincipal()
The java.security.Principal returned from this method MUST be an instance of org.eclipse.microprofile.jwt.JsonWebToken.
javax.servlet.http.HttpServletRequest#isUserInRole(String)
This method MUST return true for any name that is included in the MP-JWT "groups" claim, as well as for any role name that has been mapped to a group name in the MP-JWT "groups" claim.
javax.security.jacc.PolicyContext.getContext("javax.security.auth.Subject.container")
The javax.security.auth.Subject returned by the PolicyContext.getContext(String key) method with the standard
"javax.security.auth.Subject.container" key, MUST return a Subject that has a java.security.Principal
of type
org.eclipse.microprofile.jwt.JsonWebToken
amongst it’s set of Principal`s returned by `getPrincipals()`
. Similarly,
Subject#getPrincipals(JsonWebToken.class)
must return a set with at least one value. This means that following code
snipet must not throw an AssertionError:
Subject subject = (Subject) PolicyContext.getContext("javax.security.auth.Subject.container");
Set<? extends Principal> principalSet = subject.getPrincipals(JsonWebToken.class);
assert principalSet.size() > 0;
Mapping MP-JWT Token to Other Container APIs
For non-Java EE containers that provide access to some form of java.security.Principal
representation of an authenticated
caller, the caller principal MUST be compatible with the org.eclipse.microprofile.jwt.JsonWebToken
interface.
Signed JWT tokens
In many cases, Json Web Tokens (JWT) are created by signing a JSON representation of the token claims by following the steps described in the JSON Web Signature(JWS) specification.
The signed JWT token itself can also be encrypted (thus becoming an inner nested token). In this case it will need to be decrypted first. Please see Encrypted JWT claims and nested tokens for more information.
Verification of JWT passed to the Microservice in HTTP requests at runtime is done with the Public Key corresponding to the Private Key held by the JWT Issuer.
At the time of JWT creation, the Issuer will sign the JWT with its Private Key before passing it to the user. Upon receiving the JWT in future HTTP requests, Microservices can then use the matching Public Key to verify the JWT and trust the user information (claims) it contains.
The goal of this chapter is to detail means of passing the Public Key from the JWT Issuer to the MicroProfile JWT implementation.
Obtaining the Public Key
In practice, the Public Key is often obtained manually from the JWT Issuer and stored in or passed to the binary of the Microservice. If your public Keys do not rotate frequently, then storing them in the binary image or on disk is a realistic option for many environments. For reference, SSL/TLS Certificates to support HTTPS, which are also Public Key based, are usually configured in the JVM itself and last for up to two years.
Alternatively, Public Keys may be obtained by the Microservice at runtime, directly from the JWT Issuer via HTTPS request. MicroProfile JWT implementations are required to support this method of fetching the Public Key from the JWT Issuer via means defined here. It should be noted, however, not all JWT Issuers support downloading of the Public Key via HTTPS request.
Supported Signature Algorithms
Support for RSA RS256 and Elliptic Curve Digital Signature Algorithm (ECDSA) ES256 is required. RSA keys used for creating and verifying RS256 signatures must be of 1024 or 2048 bits in length. Other RSA key sizes are allowed, but should be considered vendor-specific.
[NOTE] Support for RSA keys of 1024 bits in length is deprecated and will become optional in the next major version of the MP JWT specification.
Other asymmetric signature algorithms are allowed, but should be considered vendor-specific. This includes Digital Signature Algorithm (DSA), Diffie-Hellman (DS), Edwards-curve Digital Signature Algorithm (EdDSA, aka ed25519).
Note
|
Symmetrically signed JWTs such as HMAC-SHA256 (hs256) are explicitly not supported, deemed insecure for a distributed Microservice architecture where JWTs are expected to be passed around freely. Use of symmetric signatures would require all microservices to share a secret, eliminating the ability to determine who created the JWT. |
Supported Public Key Formats
RSA and ECDSA Public Keys may be formatted in any of the following formats, specified in order of precedence:
-
Public Key Cryptography Standards #8 (PKCS#8) PEM
-
JSON Web Key (JWK)
-
JSON Web Key Set (JWKS)
-
JSON Web Key (JWK) Base64 URL encoded
-
JSON Web Key Set (JWKS) Base64 URL encoded
Attempts to parse the Public Key text will proceed in the order specified above until a valid Public Key can be derived.
Support for other Public Key formats such as PKCS#1, SSH2, or OpenSSH Public Key format is considered optional.
MicroProfile JWT implementations are required to throw a DeploymentException
when given
a public key that cannot be parsed using either the standardly supported or
vendor-specific key formats.
MicroProfile JWT implementations are required to throw a DeploymentException
when given
a Private Key in any format.
PKCS#8
Public Key Cryptography Standards #8 (PKCS#8) PEM format is a plain text format and is the default format for OpenSSL, many public/private key tools and is natively supported in Java.
The format consists of a Base64 URL encoded value wrapped in a standard -----BEGIN PUBLIC
KEY-----
header and footer. The Base64 URL encoded data can be decoded and the resulting
byte array passed directly to java.security.spec.PKCS8EncodedKeySpec
.
The following is an example of a valid RSA 2048 bit Public Key in PKCS#8 PEM format.
-----BEGIN PUBLIC KEY-----
MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA0440JtmhlywtkMvR6tTM
s0U6e9Ja4xXj5+q+joWdT2xCHt91Ck9+5C5WOaRTco4CPFMBxoUPi1jktW5c+Oyk
nOIACXu6grXexarFQLjsREE+dkDVrMu75f7Gb9/lC7mrVM73118wnMP2u5MOQIoX
OqqC1y1gaoJaLp/OjTiJGCm4uxzubzUPN5IDAFaTfK+QErhtcGeBDwWjvikGfUfX
+WVq74DOoggLiGbB4jsT8iVXEm53JcoEY8nVr2ygr92TuU1+xLAGisjRSYJVe7V1
tpdRG1CiyCIkqhDFfFBGhFnWlu4gKMiT0KToA9GJfOuCz67XZEAhQYizcXbn1uxa
OQIDAQAB
-----END PUBLIC KEY-----
MicroProfile JWT implementations must inspect the supplied Public Key body for the
-----BEGIN PUBLIC KEY-----
header and parse the key as PKCS#8 if found.
Support for the legacy PKCS#1 format is not required and should be considered vendor-specific.
PKCS#1 formatted keys can be identified by the use of the -----BEGIN RSA PUBLIC KEY-----
.
MicroProfile JWT implementations are required to throw a DeploymentException
if a Private Key is supplied.
JSON Web Key (JWK)
JSON Web Key (JWK) allows for a Public Key to be formatted in json and optionally Base64 encoded.
At minimum JWK formatted RSA Public Keys must contain the kty
field set to "RSA" as well as the n
and e
fields.
At minimum JWK formatted ECDSA Public Keys must contain the kty
field set to "EC", crv
field set to "P-256" as well as the x
and y
fields.
The following example is the previously shown PKCS#8 PEM formatted Public Key converted to JWK format.
{
"kty": "RSA",
"n": "sszbq1NfZap2IceUCO9rCF9ZYfHE3oU5m6Avgyxu1LmlB6rNPejO-eB7T9iIhxXCEKsGDcx4Cpo5nxnW5PSQZM_wzXg1bAOZ3O6k57EoFC108cB0hdvOiCXXKOZGrGiZuF7q5Zt1ftqIk7oK2gbItSdB7dDrR4CSJSGhsSu5mP0",
"e": "AQAB"
}
The next example shows the ECDSA Public Key:
{
"kty":"EC",
"crv":"P-256",
"x":"w4HohvwOj21FBQE1PrJOAlPRQMyWimmXH9rIHa7YMTU",
"y":"osZEjUhZa79-kClcGm79eX0q_QFLlrA99MhkzNy6MtI"
}
MicroProfile JWT implementations are required to throw a DeploymentException
if the JWK
kty
field is missing or JSON text is found, but does not follow either JWK or JWKS
format.
The JWK may be supplied in plain JSON or Base64 URL encoded JSON format.
See RFC-7517 for further details on JWK format and optional fields.
JSON Web Key Set (JWKS)
The JSON Web Key Set (JWKS) format allows for multiple keys to supplied, which can be useful for either key rotation or supporting environments that have multiple JWT Issuers and therefore multiple Public Keys.
An example of a valid JWKS:
{
"keys": [
{
"kid": "orange-1234",
"kty": "RSA",
"n": "sszbq1NfZap2IceUCO9rCF9ZYfHE3oU5m6Avgyxu1LmlB6rNPejO-eB7T9iIhxXCEKsGDcx4Cpo5nxnW5PSQZM_wzXg1bAOZ3O6k57EoFC108cB0hdvOiCXXKOZGrGiZuF7q5Zt1ftqIk7oK2gbItSdB7dDrR4CSJSGhsSu5mP0",
"e": "AQAB"
},
{
"kid": "orange-5678",
"kty": "RSA",
"n": "xC7RfPpTo7362rzATBu45Jv0updEZcr3IqymjbZRkpgTR8B19b_rS4dIficnyyU0plefkE2nJJyJbeW3Fon9BLe4_srfXtqiBKcyqINeg0GrzIqoztZBmmmdo13lELSrGP91oHL-UtCd1u5C1HoJc4bLpjUYxqOrJI4mmRC3Ksk5DV2OS1L5P4nBWIcR1oi6RQaFXy3zam3j1TbCD5urkE1CfUATFwfXfFSPTGo7shNqsgaWgy6B205l5Lq5UmMUBG0prK79ymjJemODwrB445z-lk3CTtlMN7bcQ3nC8xh-Mb2XmRB0uoU4K3kHTsofXG4dUHWJ8wGXEXgJNOPzOQ",
"e": "AQAB"
}
]
}
If the incoming JWT uses the kid
header field and there is a key in the supplied JWK set
with the same kid
, only that key is considered for verification of the JWT’s digital
signature.
For example, the following decoded JWT would involve a check on only the orange-5678
key.
{
"alg": "RS256",
"typ": "JWT",
"kid": "orange-5678"
}.
{
"sub": "1234567890",
"name": "John Doe",
"admin": true,
"iat": 1516239022
}
The JWKS may be supplied in plain JSON or Base64 URL encoded JSON format.
Signature Verification Configuration Parameters
See the Verification of JWT token claims section how to verify the token claims once its signature has been verified.
mp.jwt.verify.publickey
The mp.jwt.verify.publickey
configuration property allows the Public Verification Key text itself to be
supplied as a string. The Public Key will be parsed from the supplied string in the order
defined in section Supported Public Key Formats.
The following example shows a Base 64 URL encoded JWK passed via system property.
java -jar movieservice.jar -Dmp.jwt.verify.publickey=eyJrdHkiOiJSU0EiLCJuI\
joieEM3UmZQcFRvNzM2MnJ6QVRCdTQ1SnYwdXBkRVpjcjNJcXltamJaUmtwZ1RSOEIxOWJfclM\
0ZElmaWNueXlVMHBsZWZrRTJuSkp5SmJlVzNGb245QkxlNF9zcmZYdHFpQktjeXFJTmVnMEdye\
klxb3p0WkJtbW1kbzEzbEVMU3JHUDkxb0hMLVV0Q2QxdTVDMUhvSmM0YkxwalVZeHFPckpJNG1\
tUkMzS3NrNURWMk9TMUw1UDRuQldJY1Ixb2k2UlFhRlh5M3phbTNqMVRiQ0Q1dXJrRTFDZlVBV\
EZ3ZlhmRlNQVEdvN3NoTnFzZ2FXZ3k2QjIwNWw1THE1VW1NVUJHMHBySzc5eW1qSmVtT0R3ckI\
0NDV6LWxrM0NUdGxNTjdiY1EzbkM4eGgtTWIyWG1SQjB1b1U0SzNrSFRzb2ZYRzRkVUhXSjh3R\
1hFWGdKTk9Qek9RIiwiZSI6IkFRQUIifQo
When supplied, mp.jwt.verify.publickey
will override other standard means to supply the
Public Key such as mp.jwt.verify.publickey.location
. Vendor-specific options for
supplying the key will always take precedence.
If neither the mp.jwt.verify.publickey
nor mp.jwt.verify.publickey.location
are supplied configuration are supplied, the MP-JWT signer configuration will
default to a vendor specific behavior as was the case for MP-JWT 1.0.
MicroProfile JWT implementations are required to throw a DeploymentException
if both
mp.jwt.verify.publickey
and mp.jwt.verify.publickey.location
are supplied.
mp.jwt.verify.publickey.location
The mp.jwt.verify.publickey.location
configuration property allows for an external or internal location
of Public Verification Key to be specified. The value may be a relative path or a URL.
MicroProfile JWT implementations are required to check the path at startup or deploy time. Reloading the Public Key from the location at runtime as well as the frequency of any such reloading is beyond the scope of this specification and any such feature should be considered vendor-specific.
mp.jwt.verify.publickey.algorithm
The mp.jwt.verify.publickey.algorithm
configuration property allows for specifying which Public Key Signature Algorithm
is supported by the MP JWT endpoint. This property can be be set to either RS256
or ES256
. Default value is RS256
.
Support for the other asymmetric signature algorithms such as RS512
, ES512
and others is optional.
mp.jwt.verify.publickey.algorithm
will provide an additional hint how to read the Public Key in the PKCS#8 PEM format as both RSA and EC Public Keys in the PKCS#8 PEM format may only have a standard -----BEGIN PUBLIC KEY-----
header and footer.
It is also recommended to use this property to whitelist the token signature algorithm. For example, MP JWT implementations should only allow an mp.jwt.verify.publickey.algorithm
algorithm instead of both RS256
and ES256
when verifying a token signature.
Relative Path
Relative or non-URL paths supplied as the location are resolved in the following order:
-
new File(location)
-
Thread.currentThread().getContextClassLoader().getResource(location)
The following example shows the file orange.pem
supplied as either a file in the
Microservice’s binary or locally on disk.
java -jar movieservice.jar -Dmp.jwt.verify.publickey.location=orange.pem
Any non-URL is treated identically and may be a path inside or outside the archive.
java -jar movieservice.jar -Dmp.jwt.verify.publickey.location=/META-INF/orange.pem
Parsing of the file contents occurs as defined in Supported Public Key Formats
file:
URL Scheme
File URL paths supplied as the location allow for explicit externalization of the file via full url.
java -jar movieservice.jar -Dmp.jwt.verify.publickey.location=file:///opt/keys/orange.pem
Parsing of the file contents occurs as defined in Supported Public Key Formats
http:
URL Scheme
HTTP and HTTPS URL paths allow for the Public Key to be fetched from a remote host, which may be the JWT Issuer or some other trusted internet or intranet location.
The location supplied must respond to an HTTP GET. Parsing of the HTTP message body occurs as defined in Supported Public Key Formats
java -jar movieservice.jar -Dmp.jwt.verify.publickey.location=https://location.dev/widget/issuer
Other forms of HTTP requests and responses may be supported, but should be considered vendor-specific.
Other URL Schemes
All other locations containing a colon will be considered as URLs and be resolved using the following method:
-
new URL(location).openStream()
Thus additional vendor-specific or user-defined options can easily be added.
Example custom "smb:" location
java -jar movieservice.jar -Dmp.jwt.verify.publickey.location=smb://Host/orange.pem -Djava.protocol.handler.pkgs=org.foo
Example stub for custom "smb:" URL Handler
package org.foo.smb;
import java.io.IOException;
import java.net.URL;
import java.net.URLConnection;
import java.net.URLStreamHandler;
/**
* The smb: URL protocol handler
*/
public class Handler extends URLStreamHandler {
@Override
protected URLConnection openConnection(URL u) throws IOException {
return // your URLConnection implementation
}
}
See java.net.URL javadoc for more details.
Parsing of the InputStream
occurs as defined in Supported Public Key Formats and must
return Public Key text in one of the supported formats.
Encrypted JWT claims and nested tokens
Some claims may contain a sensitive information. For example, a user https://openid.net/specs/openid-connect-core-1_0.html#AddressClaim [Address claim] can be easily viewed if signed JWT token containing such a claim is intercepted or leaked in the logs.
In such cases, when the confidentiality of the claims is critical, the claims can be encrypted or the signed JWT can be encrypted, producing a JWT token by following the steps in the JSON Web Encryption(JWE) specification.
Decrypting the encrypted claims requires a single property, mp.jwt.decrypt.key.location
, pointing to a private key which can be used to decrypt the token. All the location options supported by the mp.jwt.verify.publickey.location
property are also supported by mp.jwt.decrypt.key.location
.
An extra care is required to ensure the private keys are not leaked, particularly, making them available at the insecure HTTP locations or as part of the application archive is not recommended. This is also why the inlined private keys are not supported.
Note that two types of keys are required to implement a JWE encryption scheme:
-
Content encryption key - typically a generated secret key which is used to encrypt a plaintext such as a JSON representation of the token claims.
-
Key management key - public RSA key which is used to encrypt a content encryption key.
mp.jwt.decrypt.key.location
must point to a private RSA key matching this key.
Key management key algorithm which must be supported is RSA-OAEP (RSAES using Optimal Asymmetric Encryption Padding) with a key length 2048 bits or higher.
Content encryption algorithm which must be supported is A256GCM (AES in Galois/Counter Mode (GCM)).
Similarly to the signature verification keys, key management keys may be formatted in any of the following formats, specified in order of precedence:
-
Private Key Cryptography Standards #8 (PKCS#8) PEM
-
JSON Web Key (JWK)
-
JSON Web Key Set (JWKS)
-
JSON Web Key (JWK) Base64 URL encoded
-
JSON Web Key Set (JWKS) Base64 URL encoded
The rules about matching a token kid
header and JWK kid
property for selecting the verification keys apply when a key management key is formatted as JWK.
If the claims have been immediately encrypted, without being signed first, then the application endpoints will have no guarantee that a token came from a trusted issuer. To have this proof the claims will need to be signed first and the resulting nested JWT token - encrypted next.
If the encrypted content is an inner nested JWT then the implementations must check that the cty
(content type) JWE header is set to JWT
and verify the signature of the nested JWT by configuring the verification key as described in the Signed JWT tokens section.
See the Verification of JWT token claims section how to verify the token claims once the token has been decrypted and the signature of its nested token (if present) verified.
Decryption Configuration Parameters
mp.jwt.decrypt.key.location
The mp.jwt.decrypt.key.location
config property allows for an external or internal location
of Private Decryption Key to be specified. The value may be a relative path or a URL.
Please see mp.jwt.publickey.location for all the information about the supported locations and Encrypted JWT claims and nested tokens section for the additional recommendations.
Verification of JWT token claims
MP JWT specification currently supports the verification of the token iss
issuer and aud
audience claims which is done after the token signature has been verified or the token has been decrypted.
mp.jwt.verify.issuer
The mp.jwt.verify.issuer
config property allows for the expected value of the iss
claim to be specified. A MicroProfile JWT implementation must verify the iss
claim of incoming JWTs is present and matches the configured value of mp.jwt.verify.issuer
.
Note that since this property verifies the iss
claim value, it will be effective irrespectively of how the token claims have been protected (signed or encrypted or signed first and encrypted next).
mp.jwt.verify.audiences
The mp.jwt.verify.audiences
config property is a comma delimited list of allowable values for the aud
claim. If specified, a MicroProfile JWT implementation must verify the aud
claim of incoming JWTs is present and at least one value in the claim matches one of the configured values of mp.jwt.verify.audiences
.
Requirements for accepting signed and encrypted tokens
MP JWT specification currently requires that an MP JWT application accepts only signed or only encrypted or only signed and encrypted tokens as it expected that many endpoints will have the requirements to accept a single token type only.
If only mp.jwt.verify.publickey.location
or mp.jwt.verify.publickey
properties are set then only the tokens containing the signed claims can be accepted. Support for such tokens is required
.
If mp.jwt.decrypt.key.location
and either mp.jwt.verify.publickey.location
or mp.jwt.verify.publickey
are set then only the tokens which contain the claims signed first and then encrypted can be accepted. Support for such tokens is required
.
If only mp.jwt.decrypt.key.location
property is set then only the tokens containing the encrypted claims can be accepted.
Support for such tokens is optional
however it is recommended
that the encrypted-only tokens are supported if they are provided to the MP JWT endpoints as cookies.
JWT and HTTP headers
Configuration Properties
mp.jwt.token.header
The mp.jwt.token.header
configuration property allows to set up the header which is expected to contain a JWT token.
MP JWT implementations are required to support Authorization
(default) or Cookie
configuration values.
Support for other headers or alternative authentication schemes is optional.
mp.jwt.token.cookie
The mp.jwt.token.cookie
configuration property allows to set up the Cookie name (default is Bearer
) which is
expected to contain a JWT token.
This configuration will be ignored unless mp.jwt.token.header
is set to Cookie
.
Providing the recommendations on how to secure a JWT token as a Cookie is out of scope for the MP JWT specification. Generally one should avoid putting sensitive user information into a signed JWT token.
How to provide Configuration Parameters
MicroProfile JWT leverages the MicroProfile Config specification to provide a consistent means of passing all supported configuration options.
Prior to MicroProfile JWT 1.1 all configuration options for the Public Key and claim verification were vendor-specific. Any equivalent vendor-specific methods of configuration are still valid and shall be considered to override any standard configuration mechanisms.
MP JWT specification allows at minimum configuration options to be specified in the microservice binary itself or via command-line via -D properties as follows:
java -jar movieservice.jar -Dmp.jwt.verify.publickey.location=orange.pem
By convention of the MicroProfile JWT specification, property names are always lowercase
and begin with mp.jwt.
Mapping Configuration Parameters to Environment Variables
When using environment variables to specify the MP-JWT configuration properties defined in this section,
note that some operating systems allow only alphabetic characters and underscores in environment variables.
Since characters such as '.' may be disallowed, in order to set a value for a config property such as mp.jwt.verify.publickey
using an environment variable, the following mapping rules from the MP configuration spec are relevant:
When searching environment variables for configuration properties, the following transformation is applied to attempt to find a match:
-
Exact match (i.e. mp.jwt.verify.publickey)
-
Replace nonalphanumeric characters with '_' (i.e. mp_jwt_verify_publickey)
-
Replace nonalphanumeric characters with '_' and convert to uppercase (i.e. MP_JWT_VERIFY_PUBLICKEY)
With these rules, the matching portable environment variables names for the current MP-JWT verification properties are:
mp.jwt.verify.publickey
-
mp_jwt_verify_publickey or MP_JWT_VERIFY_PUBLICKEY
mp.jwt.verify.publickey.location
-
mp_jwt_verify_publickey_location or MP_JWT_VERIFY_PUBLICKEY_LOCATION
mp.jwt.verify.publickey.algorithm
-
mp_jwt_verify_publickey_algorithm or MP_JWT_VERIFY_PUBLICKEY_ALGORITHM
mp.jwt.verify.issuer
-
mp_jwt_verify_issuer or MP_JWT_VERIFY_ISSUER
mp.jwt.verify.audiences
-
mp_jwt_verify_audiences or MP_JWT_VERIFY_AUDIENCES
mp.jwt.token.header
-
mp_jwt_token_header or MP_JWT_TOKEN_HEADER
mp.jwt.token.cookie
-
mp_jwet_token_cookie or MP_JWT_TOKEN_COOKIE
mp.jwt.decrypt.key.location
-
mp_jwt_decrypt_key_location or MP_JWT_DECRYPT_KEY_LOCATION
Future Directions
Not all considerations discussed during the specification process make it into the specification. This section serves as an abridged version for the purposes of soliciting feedback and interest. By convention we will leave items in the Future Direction for at most two revisions.
resource_access
claim
In future versions of the API we would like to address service specific group claims. The resource_access
claim originally targeted for the 1.0 release of the specification has been postponed as additional work to determine
the format of the claim key as well as how these claims would be surface through the standard Java EE APIs needs
more work than the 1.0 release timeline will allow.
For reference, a somewhat related extension to the OAuth 2.0 spec Resource Indicators for OAuth 2.0 has not gained much traction. The key point it makes that seems relevant to our investigation is that the service specific grants most likely need to be specified using URIs for the service endpoints. How these endpoints map to deployment virtual hosts and partial, wildcard, etc. URIs needs to be determined.
roles
claim
A roles
claim was considered in addition to the groups
claim that made it into the final
specification. The groups
claim is intended to be mapped to specific roles on the target resource server. The roles
claim was intended to explicitly state roles inside the JWT that would not be subject to any mappings and are made
available to directly to the application for @RolesAllowed
and similar RBAC checks. The roles in the JWT should
be exposed to the app in addition to any roles that result from group-to-role mapping provided by the target resource server.
The intended JWT datatype for roles
should be a JSON string array.
Though a roles
claim is not required, implementations that support it and applications that use it should do so as
detailed in this section to ensure alignment for any future standardization.
aud
claim
The aud
claim defined in RFC 7519 section 4.1.3 was considered for addition. The intended JWT datatype for
aud
should be a JSON string array or a single string as defined in RFC 7519.
Though an aud
claim is not required, implementations that support it and applications that use it should do so as
detailed in this section to ensure alignment for any future standardization.
Enabling/Disabling Claim Requirements
We discussed adding configuration settings to disable/enable the requirements for
claims such as the iss
claim, but could not decide on the format for such
configuration given the time constraints of the 1.1 release. We will revisit
this issue in the next release.
classpath:
URL Scheme
The option to have a built-in classpath:
URL Scheme was discussed with the intended
benefit of providing some way to explicitly state a Public Key file is inside the archive
and to remove potential a similarly named file existed on disk.
For the moment this was deemed an edge-case that could be solved with a custom URL Scheme and consensus that this would add to the complexity of the specification. In practice, this may be very useful so those who find themselves with this scenario are encouraged to contact the MicroProfile discussion lists.
Expiration tolerance
Relaxing or potentially ignoring the expiration time of a JWT was discussed and deemed an attractive option for future standardization. It was omitted in efforts to keep the first revision of the configuration as simple as possible.
Users who find themselves with this need are encouraged to both request support from their respective implementation and to detail their use case the MicroProfile discussion lists, so any future standardization work accounts for all scenarios.
Passing JWTs as Cookies
Semantics for passing JWTs via HTTP Cookie
headers instead of the HTTP Authorization
header
were discussed and deemed a valuable addition for interoperability purposes.
It is likely to be a future recommendation that MicroProfile JWT implementation supporting the
transport of MP-JWT tokens using cookies SHOULD recognize the cookie name Bearer
, and the cookie value format being the JWS Compact Serialization as described in section 7.1 of JSON Web Signature (RFC-7515).
There are a number of security considerations when using cookies to transfer security information such as tokens in cookies. Issues related to CSRF, XSS, and storage need to be considered. The following links present material on these issues:
The motivation for using Bearer
as the Cookie name is to establish a pattern of using the Authorization
scheme as the cookie name. Bearer token example would look as follows:
-
Authorization: Bearer as14efgscd31qrewtadg
-
Cookie: Bearer=as14efgscd31qrewtadg
Basic auth would look as follows:
-
Authorization: Basic gasdqwe198abg313ffd
-
Cookie: Basic=gasdqwe198abg313ffd
Sample Implementations
This sections references known sample implementations of the Eclise MicroProfile JWT RBAC authorization specification.
General Java EE/SE based Implementations
A baseline sample implementation that is available under the Apache License, Version 2.0 can be found at https://github.com/MicroProfileJWT/microprofile-jwt-auth-prototype. The purpose of this prototype is to offer reusable code for integration of the JWT RBAC authenticaiton and authorization spec in various container environments. This particular implementation contains:
-
a default implementation of the JsonWebToken interface
-
a JAX-RS ContainerRequestFilter prototype
-
JSR-375 IdentityStore and Credential prototypes
Wildfly Swarm Implementations
A sample implementation for a custom auth-method of MP-JWT with the Wildfly/Wildfly-Swarm Undertow web container, as a Wildfly-Swarm fraction, available under the Apache License, Version 2.0 can be found at: https://github.com/MicroProfileJWT/wfswarm-jwt-auth-fraction.
Release Notes for MicroProfile JWT 1.2
A full list of changes delivered in the 1.2 release can be found at MicroProfile JWT 1.2 Milestone.
API Changes
-
A convenience method has been added to allow retrieving claims from
JsonWebToken
by using theClaims
enum (#154)
Spec Changes
-
Support for verifying JWT tokens which have been signed using Elliptic Curve
ES256
signature algorithm (#161) -
Support for decrypting JWT tokens which have been encrypted using
RSA-OAEP
andA256GCM
algorithms and contain the claims or inner-signed JWT tokens (#58) -
Support for JWT audience
aud
claim (#121) -
Support for JWT token cookies (#93)
-
JWT token
groups
claim is now optional (#129) -
Better specification of the injection point (#116, #127), scope (#45, #183) and required claims (#128) requirements
-
Support for RSA keys of 1024 bit length has been deprecated (#197)
Release Notes for MicroProfile JWT 1.1.1
Changes in 1.1.1-RC2
This candidate patch release provides one additional TCK test fix, and information on how to use the MP-JWT configuration properties as environment variables on platforms that don’t allow dots '.' in their names.
There are no API changes.
Changes in 1.1.1-RC1
This is a candidate patch release that provides fixes to the TCK tests that were seen to have issues in different implementations. The TCK tests WARs now include a META-INF/MPJWTTESTVERSION resource that contains the major/minor version string enum for the MP-JWT version the test WAR is targeting. Currently the versions enums are:
public enum MpJwtTestVersion {
MPJWT_V_1_0,
MPJWT_V_1_1
;
}
There are no API changes.
Release Notes for MicroProfile JWT 1.1
The focus of this release was to add support for configuring the public key and issuer needed for verification of the MP-JWT using MicroProfile Config. The new MicroProfile Config properties are:
- mp.jwt.verify.publickey
-
The embedded key material of the public key for the MP-JWT signer in PKCS8 PEM or JWK(S) format. If not found the mp.jwt.verify.publickey.location needs to be checked.
- mp.jwt.verify.publickey.location
-
The relative path or full URL of the public key. All relative paths will be resolved within the archive using ClassLoader.getResource. If the value is a URL it will be resolved using new URL(“…”).openStream()
- mp.jwt.verify.issuer
-
The expected iss claim value to validate against an MP-JWT.