Implementation:Ray project Ray Ray Shutdown

Knowledge Sources	Ray
Domains	Distributed_Computing, Resource_Management
Last Updated	2026-02-13 17:00 GMT

Overview

Concrete tool for gracefully shutting down the Ray distributed runtime provided by the Ray Java SDK.

Description

The Ray.shutdown() static method is a synchronized method that calls runtime.shutdown() on the internal RayRuntime instance and sets the static reference to null. It disconnects from the cluster scheduler (Raylet) and Global Control Store (GCS), releases all native JNI resources, and prevents further distributed operations. The method is safe to call multiple times (idempotent).

Usage

Call Ray.shutdown() when your application is done using Ray. It is also registered as a JVM shutdown hook during Ray.init(), so it runs automatically on JVM exit if not called explicitly.

Code Reference

Source Location

• Repository: ray-project/ray
• File: java/api/src/main/java/io/ray/api/Ray.java
• Lines: L37-42

Signature

public static synchronized void shutdown()

Import

import io.ray.api.Ray;

I/O Contract

Inputs

Name	Type	Required	Description
(none)	—	—	Zero-argument method. Operates on the static Ray.runtime field.

Outputs

Name	Type	Description
Ray.runtime	null	Static field set to null after shutdown

Usage Examples

Explicit Shutdown

import io.ray.api.Ray;

public class ShutdownExample {
    public static void main(String[] args) {
        Ray.init();

        // ... perform distributed operations ...

        // Explicit graceful shutdown
        Ray.shutdown();
        // Ray.runtime is now null; any further Ray calls will throw IllegalStateException
    }
}

Related Pages

Implements Principle

• Principle:Ray_project_Ray_Runtime_Shutdown

Requires Environment

• Environment:Ray_project_Ray_Java_Build_Environment