Discovery Feed Handler
Identified by type="DiscoveryFeed"
, this handler produces a combined JSON feed of IdP information from all <MetadataProvider>
 sources that support the feature.
While producing this information through the SP is not ideal from an efficiency point of view, doing so helps ensure accurate information is supplied and addresses synchronization issues across the metadata sources as they internally refresh themselves. Deployers needing additional performance may choose to manually access the handler to produce a static file for delivery, updating it on some schedule.
Attributes
Common Attributes
Specific Attributes
Name | Type | Default | Description |
---|---|---|---|
cacheToClient | boolean | false | If true, the feed includes cache directives intended to support client-side caching of the information. Many clients are overly aggressive about caching, so this feature is disabled by default. |
cacheToDisk | boolean | true | If true, discovery feeds are cached to disk by the shibd process, and read back from disk while producing the feed from the web server. This is much more efficient, but requires shared file access. If a shibd process is used remotely, this can be disabled to supply the feed through the remoting layer |
dir | relative or absolute path | If set and cacheToDisk is true, specifies the directory to store feed files. If omitted, feeds will be written to the package-specific "run" directory, usually /var/run/shibboleth or /opt/shibboleth-sp/var/run/shibboleth |