Researcher REST API¶
Researcher Kubernetes API is deprecated. See Cluster API for its replacement.
The purpose of the Researcher REST API is to provide an easy-to-use programming interface for submitting, listing, and deleting Jobs.
There are other APIs that provide the same functionality. Specifically:
- If your code is script-based, you may consider using the Run:ai command-line interface.
- You can communicate directly with the underlying Kubernetes infrastructure by sending YAML files or by using a variety of programming languages to send requests to Kubernetes. See Submit a Run:ai Job via Kubernetes API.
The Researcher REST API is cluster-specific in the sense that if you have multiple GPU clusters, you will have a separate URL per cluster. This
<CLUSTER-ENDPOINT> can be found in the Run:ai User Interface, under
Clusters. Each cluster will have a separate URL.
- By default, researcher APIs are unauthenticated. To protect researcher API, you must configure researcher authentication.
- Once configured, you must create a Client Application to make API requests. Use the client application and secret, to obtain a time-bound bearer token (
<ACCESS-TOKEN>). For details, see Calling REST APIs.
- Use the token for subsequent API calls.
Get all the jobs for a project named
Researcher API Scope¶
The Researcher API provides the following functionality:
- Submit a new Job
- List jobs for specific Projects.
- Delete an existing Job
- Get a list of Projects for which you have access to
Researcher API Documentation¶
To review API documentation:
- Open the Run:ai user interface
- Go to
- Locate your cluster and browse to
- When using the
Bearer <ACCESS TOKEN>(simply adding the access token will not work).
The document uses the Open API specification to describe the API. You can test the API within the document after creating and saving a token.