The Shibboleth IdP generally requires SAML metadata to provision connectivity with SAML relying parties and inform it about their capabilities and technical specifics. While you have the option to operate in a more "promiscuous" way (by enabling profiles for "unverified" RPs), this is not a common operating mode. In most cases, you will configure metadata sources in order to use the IdP's SAML features; this is done by adding <MetadataProvider> elements inside the metadata-providers.xml file.
Note that the other protocol(s) supported by the IdP generally support, but not strictly require, the use of SAML metadata, alongside other protocol-specific means of handling the same kinds of functions. Use of metadata enables a variety of useful configuration techniques that are superior to the features available from other approaches to managing relying parties and is generally advisable.
A typical use case is to load (and periodically reload) entity metadata from a local file:
Another use case is to load (and periodically reload) a metadata aggregate from a remote source via HTTP:
But increasingly, the dynamic providers (LocalDynamicMetadataProvider and DynamicHTTPMetadataProvider) are used in lieu of the reloading providers (FilesystemMetadataProvider and FileBackedHTTPMetadataProvider) shown above. See the MetadataManagementBestPractices topic for use cases and recommendations.
The ChainingMetadataProvider is often used to combine two or more metadata sources. The metadata-providers.xml file that ships with the software includes such a chain "wrapper" by default.
Namespaces and Schemas
Nearly all elements described in this page and its children are defined in the urn:mace:shibboleth:2.0:metadata namespace, the schema for which can be located at http://shibboleth.net/schema/idp/shibboleth-metadata.xsd. Throughout this document and its children, this is assumed to be the default XML namespace in effect. The namespace prefix "metadata:" is conventionally also bound to this namespace.
The precise behavior of any <MetadataProvider> element is controlled by the xsi:type attribute (see below). The following types are supported and examples are provided for each type. If the urn:mace:shibboleth:2.0:metadata namespace is not the default, then a prefix (presumably "metadata:") is required when specifying these types.
A provider that allows metadata to be specified inline
Configuration options common to two or more metadata providers are listed in the subsections below. Others are specific to the xsi:type, and these are documented on the pages specific to each type.
As described in the ReloadableServices documentation, the configuration is actually loaded from a bean whose name is specified by the property idp.service.metadata.resources, with the default value shibboleth.MetadataResolverResources (in turn defined in conf/services.xml to be a list with one entry: the file metadata-providers.xml)
You can, if you choose, override this with additional or different files or more advanced sources. Each resource usually is expected to supply a "top level" <MetadataProvider> element with attributes and child elements as described above. Search order amongst multiple top level elements is arbitrated by the sortKey attribute, where lower values are processed before higher ones.
It is also now possible to declare "by reference" metadata filters that are attached by reference, so it is possible to supply a resource whose top-level element is such a <MetadataFilter> in such cases.
If a specific relying party (as identified by a specific entityID) is duplicated in the metadata sources provided, then which precise entry is chosen is governed by the following rules:
Metadata sources combined via a chain are searched in the order in which they occur in the chain, and the first entry matching the entityID is returned.
If multiple "top level" Metadata Providers are provided then they are searched in an order derived from the (numeric) value of the sortKey attribute (lowest key first). If no sortKey is specified, then the search order is undefined.
In whatever order of sources is in effect, the first entry matching the entityID is returned.
If a single metadata source contains multiple entries with the same entityID, then which entry is returned is undefined (exception: invalid entries would be ignored in favor of valid ones in most cases).