Implementation:Ray project Ray Serve Run
| Knowledge Sources | |
|---|---|
| Domains | Model_Serving, Deployment_Orchestration |
| Last Updated | 2026-02-13 17:00 GMT |
Overview
Concrete tool for deploying application DAGs to the Ray Serve control plane provided by the Ray Java Serve SDK.
Description
Serve.run() takes an Application from .bind(), calls Graph.build() to transform the DAG into a flat list of Deployment objects, serializes them to protobuf, and sends them to the Python Serve controller via ServeControllerClient.deployApplication(). When blocking=true (default), it waits for all replicas to reach RUNNING state, then returns a DeploymentHandle.
Usage
Call Serve.run() after binding a deployment to deploy it to the cluster.
Code Reference
Source Location
- Repository: ray-project/ray
- File: java/serve/src/main/java/io/ray/serve/api/Serve.java (L310-370)
- File: java/serve/src/main/java/io/ray/serve/dag/Graph.java (L15-19)
Signature
public static DeploymentHandle run(Application application)
public static DeploymentHandle run(Application application, String name, String routePrefix, boolean blocking)
Import
import io.ray.serve.api.Serve;
import io.ray.serve.deployment.Application;
import io.ray.serve.handle.DeploymentHandle;
I/O Contract
Inputs
| Name | Type | Required | Description |
|---|---|---|---|
| application | Application | Yes | The bound deployment from .bind() |
| name | String | No | Application name (default: "default") |
| routePrefix | String | No | HTTP route prefix |
| blocking | boolean | No | Wait for RUNNING state (default: true) |
Outputs
| Name | Type | Description |
|---|---|---|
| handle | DeploymentHandle | Handle for sending requests to the deployed application |
Usage Examples
Deploy and Get Handle
import io.ray.serve.api.Serve;
import io.ray.serve.deployment.Application;
import io.ray.serve.handle.DeploymentHandle;
Serve.start(new java.util.HashMap<>());
Application app = Serve.deployment()
.setName("my-service")
.setDeploymentDef("com.example.MyHandler")
.setNumReplicas(2)
.bind();
// Deploy and wait for replicas to be RUNNING
DeploymentHandle handle = Serve.run(app);
// Send requests via handle
Object result = handle.remote("input-data").result();
Serve.shutdown();
Related Pages
Implements Principle
Requires Environment
Uses Heuristic
Page Connections
Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment