Finding services#

Repertoire’s discovery endpoint provides the URLs of other services running in the same Phalanx environment.

Service types#

Services are divided into three classes:

  1. Data services: REST APIs, one base URL per dataset name (although for some services all the URLs may be the same), and intended for use either by users directly or by other services.

  2. UI services: browser UIs, one entry-point URL per Phalanx environment, intended for use by users or administrators.

  3. Internal services: REST APIs, one base URL per Phalanx environment, used primarily (but not necessarily exclusively) by other services and not by users directly.

Service APIs#

DiscoveryClient provides the following methods to look up services:

DiscoveryClient.url_for_data_service

Takes the name of the service and the name of the dataset and returns the corresponding base URL for that service’s REST API, or None if that service is not running in the local Phalanx environment or if it doesn’t provide that dataset. See Listing datasets for information on how to find the dataset names available in the local environment.

DiscoveryClient.url_for_internal_service

Takes the name of the service and returns the base URL for that service’s REST API, or None if that service is not running in the local Phalanx environment.

DiscoveryClient.url_for_ui_service

Takes the name of the service and returns the entry-point URL for the browser-based user interface, or None if that service is not running in the local Phalanx environment.

URLs are returned as strings. These APIs do not require authentication.

Butler configuration#

Configuring a client of the Butler service is a special case. The base URL of the Butler service REST API is not available through the above service APIs. The Butler client instead requires a URL to a configuration file.

Clients that need to instantiate a single Butler for a specific dataset should call DiscoveryClient.butler_config_for and pass in the name of a dataset. The return value will be a URL to the Butler configuration for that dataset, or None if no Butler for that dataset is available. This URL, in turn, can be used to the Butler constructor.

Clients that need to handle all available datasets and create a Butler on the fly for each request should call DiscoveryClient.butler_repositories and pass the result into the constructor of LabeledButlerFactory. The resulting factory can then be used to create Butler instances for labels as needed.

Next steps#