Atlassian uses cookies to improve your browsing experience, perform analytics and research, and conduct advertising. Accept all cookies to indicate that you agree to our use of cookies on your device. Atlassian cookies and tracking notice, (opens new window)
The Shibboleth IdP V4 software has reached its End of Life and is no longer supported. This documentation is available for historical purposes only. See the IDP5 wiki space for current documentation on the supported version.
The most typical options used are described in more detail below, but not every obscure option is discussed. See the javadoc for all of the possible configuration options for this profile (note that many of them are inherited from parent classes).
Virtually all the configuration options below can be set via two different properties: a static property that explicitly sets the value to use and a lookup strategy or predicate property that takes a Function or Predicate and returns the value to use. The dynamic property is generally named "propertyNamePredicate" or "propertyNameLookupStrategy" for Boolean- and non-Boolean-valued properties respectively.
Bean named shibboleth.DefaultSecurityConfiguration
An object containing all of the default security-related objects needed for peer authentication and encryption. See SecurityConfiguration for complete details.
disallowedFeatures
Integer
0
A bitmask of features to disallow, the mask values being specific to individual profiles
inboundInterceptorFlows
List<String>
Ordered list of profile interceptor flows to run prior to message processing
outboundInterceptorFlows
List<String>
Ordered list of profile interceptor flows to run prior to outbound message handling
specify an alternate signing or decryption key to use
control signing or encryption algorithms (but for metadata you control, it's advisable to control algorithms by using an extension to specify supported algorithms).
The two interceptor lists allow the much less commonly used profile interceptor injection points to be used. This is largely a Java-based way of doing very low-level sorts of “message rewriting” hackery that might otherwise be impossible to pull off. One use case for the inbound side might be picking up non-standard parameters in a SAML request.
Options common to SAML profiles:
Name
Type
Default
Description
Name
Type
Default
Description
signResponses
Boolean
varies by profile
Whether to sign responses
signRequests
Boolean
false
Whether to sign requests
Guidance
It isn't too common to need any of these options, and they should be changed only with care.
The signResponses default varies by profile, see the notes on the individual profile pages.
Options common to SAML 2.0 profiles:
Name
Type
Default
Description
Name
Type
Default
Description
ignoreRequestSignatures
Boolean
false
Whether to skip validation of signatures on requests
encryptionOptional
Boolean
false
Whether to automatically disable encryption if the relying party does not possess a suitable key
encryptNameIDs
Boolean
varies by profile
Whether to encrypt NameIDs
Guidance
The encryption options are generally set correctly for each different profile; see the notes on the individual profile pages.
Note that when the conditions to encrypt various constructs evaluate to true, the IdP will fail the request if it is unable to perform the encryption, for whatever reason. This is overrideable using the encryptionOptional property, which allows the IdP to encrypt if it can but continue otherwise. If you carefully control your metadata sources, which you should do in any case, you should be able to trust that any SP lacking an encryption key is incapable of encryption anyway, making the property safe to enable.
The ignoreRequestSignatures option is an interoperability knob to deal with badly broken or incompetently operated services. Signed requests in some profiles, particularly SSO, are often pointless and are frequently used for no good reason. If the signer's code is broken, or even worse if they manage their key poorly and require constant flag days to update them, this allows the signature to be ignored and potentially the key to be bypassed so their incompetence doesn't impact your operations.
Options common to SAML profiles that create assertions:
Name
Type
Default
Description
Name
Type
Default
Description
additionalAudiencesForAssertion
Collection<String>
Additional values to populate into audience restriction condition of assertions
includeConditionsNotBefore
Boolean
true
Whether to include a NotBefore attribute in assertions
assertionLifetime
Duration
PT5M
Lifetime of assertions
signAssertions
Boolean
false
Whether to sign assertions
Guidance
It isn't too common to need any of these options, and they should be changed only with care.
The additionalAuduencesForAssertion and includeConditionsNotBefore settings provide ways to work around bugs in other systems. You should never use these settings without obtaining a commitment from the other system's owner to fix their bugs.
The assertionLifetime setting does not involve control over the session with the relying party, it's only relevant in delegation scenarios.
The signResponses default varies by profile, see the notes on the individual profile pages.
If you need to enable the signAssertions option, and you control the SP's metadata, you should generally add the WantAssertionsSigned flag to it in place of using this option. Related, the idp.saml.honorWantAssertionsSigned property can be turned off to globally ignore that flag in metadata should you wish to do so.
Options specific to the SAML 2.0 Browser SSO profile:
Name / Type
Default
Description
Name / Type
Default
Description
encryptAssertions
Boolean
true
Whether to encrypt assertions as a whole
encryptAttributes
Boolean
false
Whether to encrypt individual SAML Attributes
maximumSPSessionLifetime
Duration
0
If non-zero, attempts to limit length of session with SP via SessionNotOnOrAfter attribute
skipEndpointValidationWhenSigned
Boolean
false
Whether to skip validation of response location via metadata if the request was signed
nameIDFormatPrecedence
List<String>
Ordered list of NameID Format(s) to select for use, in the event that a relying party does not signal a preference.
ignoreScoping
Boolean
false
Whether to ignore <saml2:Scoping> elements within an SP's AuthnRequest, and bypass generating one in accordance with the standard when proxying
checkAddress
Boolean
true
Whether to enforce consistency between the client's address and the value within an inbound assertion's <saml2:SubjectConfirmationData> and <saml2:SubjectLocality> elements
proxyCount
Non-negative Integer
Controls the insertion of a proxy count into a <saml2:Scoping> element (when issuing SAML 2 AuthnRequests to an IdP) and into a <saml2:ProxyRestiction> element (when issuing SAML 2 assertions)
proxyAudiences
Set<String>
Controls the insertion of audiences into a <saml2:ProxyRestiction> element when issuing SAML 2 assertions
proxiedAuthnInstant
Boolean
true
Whether to pass through a proxied AuthnInstant value from an inbound assertion when issuing new assertions based on it (the alternative is to insert a fresh timestamp)
suppressAuthenticatingAuthority 4.2
Boolean
false
Whether to prevent the insertion of <AuthenticatingAuthority> elements(s) in the event of proxying
maximumTimeSinceAuthn
Duration
Limits the allowable time to accept a proxied authentication assertion based on its AuthnInstant, this is principally used to cross-check use of the ForceAuthn flag
authnContextComparison
"exact", "minimum", "maximum", "better"
see below
Controls the comparison operator used when including <saml2p:RequestedAuthnContext> elements in proxied AuthnRequests
Controls bidirectional translation of <saml2:AuthnContext> content when issuing requests and generating assertions to allow for remapping of values across the proxy boundary
More advanced support for populating <saml2:AuthnContext> content based on arbitrary request state (e.g. use of SAML Attributes from a proxied IdP)
requireSignedRequests 4.3
false
When true, equivalent to setting the AuthnRequestsSigned attribute in SP metadata, blocks unsigned requests. Main use for this is to facilitate blocking IdP-initiated SSO.
Guidance
The nameIDFormatPrecedence property is a common way of controlling the type of SAML NameIdentifier / NameID included in a response, a common requirement of many commercial services. It is in fact the only way to force the use of the ill-advised "urn:oasis:names:tc:SAML:1.1:nameid-format:unspecified" Format, which it must be noted is very rarely needed, despite frequent mis-documentation to the contrary.
In most cases, it is better to control the Format selected by including a <NameIDFormat> element in the SP's metadata. In the event that you don't control the metadata, you can inject the required element by applying a metadata filter.
Examples of nameIDFormatPrecedence property
<bean id="shibboleth.DefaultRelyingParty" parent="RelyingParty">
<!--
Both constants below evaluate to the string "urn:oasis:names:tc:SAML:1.1:nameid-format:unspecified",
and are interchangeable, they're just illustrations of different ways to reference the same string.
-->
<!-- Use "unspecified" NameIdentifier with Shibboleth SSO profile. -->
<bean parent="Shibboleth.SSO">
<property name="nameIDFormatPrecedence">
<list>
<util:constant static-field="org.opensaml.saml.saml1.core.NameIdentifier.UNSPECIFIED" />
</list>
</property>
</bean>
<!-- Use "unspecified" NameID with SAML 2 SSO profile. -->
<bean parent="SAML2.SSO">
<property name="nameIDFormatPrecedence">
<list>
<util:constant static-field="org.opensaml.saml.saml2.core.NameIDType.UNSPECIFIED" />
</list>
</property>
</bean>
<!-- Return formats from a function bean (not shown). -->
<bean parent="Shibboleth.SSO" p:nameIDFormatPrecedenceLookupStrategy-ref="FormatsFunction" />
</bean>
The skipEndpointValidationWhenSigned option is attractive in many enterprise scenarios if you prefer to maintain some degree of security but avoid registration of metadata containing every individual SP endpoint, which adds a lot of overhead in massively vhosted-environments. It can also add a degree of insulation from SP changes, but in practice systems that are likely to change endpoint locations but don't support metadata-based change control are likely to misunderstand the need to keep entityIDs stable also.
The ignoreScoping setting is provided to work around interoperability issues with broken SPs.
Several other new settings are used when proxying and provide various kinds of policy controls familiar to SP operators, as well as new features to support remapping of potentially non-interoperable AuthnContext values. By default, the IdP operates in a fairly automatic fashion when proxying, such that any <saml2p:RequestedAuthnContext> element from an SP will be echoed essentially as-is to any upstream Identity Provider, and the data found in the incoming assertion will be echoed as-is back downstream. Since proxying is often used to firewall against interoperability problems and crosswalk between different communities of practice, functions can be plugged in to perform more flexible mapping of values, and some pre-existing machinery is in place to support this declaratively, as described in the AuthenticationConfiguration page. An additional hook was added in V4.1 that allows a similar function to be injected but with access to the entire request state to do more advanced things.
There are also a variety of settings related to delegation that are not shown above but can be found in the relevant API documentation.
ECP-specific settings:
Name
Type
Description
Name
Type
Description
localEvents
Collection<String>
Identifies webflow events that should be handled without use of a SOAP fault
Notes
The default value of signResponses for this profile is "true", in keeping with modern best practice. As long as one of the response or assertion are signed, use of the profile is "safe" in terms of authentication integrity, but there are vulnerabilities in XML Encryption that make signing responses advisable when the most common encryption algorithms are used. Some of the backstory around the signing defaults is discussed in this thread.
The default value of encryptAssertions for this profile is "true".
If you encounter a relying party that accepts an unsigned response and assertion that is transmitted via the ECP profile, you have identified an insecure implementation and should report the issue immediately while following your local security incident response process.