Implementation:Apache Shardingsphere KillProcessHandler Handle

Knowledge Sources	Apache_Shardingsphere
Domains	Cluster, Process_Management
Last Updated	2026-02-10 00:00 GMT

Overview

KillProcessHandler handles distributed kill process events by extracting instance and process IDs from node paths and delegating to ProcessRegistry for execution.

Description

KillProcessHandler implements GlobalDataChangedEventHandler and subscribes to KillProcessTriggerNodePath for ADDED and DELETED event types. On an ADDED event, it extracts the instance ID and process ID from the event path using NodePathSearcher. If the instance ID matches the local instance, it calls ProcessRegistry.kill(processId) to terminate the SQL process and then cleans up the trigger path via ClusterProcessPersistCoordinator. On a DELETED event, it notifies the ProcessOperationLockRegistry to release any waiting locks for the process ID. SQL exceptions during kill are wrapped in SQLWrapperException.

Usage

This handler is loaded via SPI as a GlobalDataChangedEventHandler in cluster mode. It is triggered when a KILL PROCESS command is issued on any cluster node, which persists a trigger path that all nodes watch. Only the node whose instance ID matches will execute the actual kill.

Code Reference

Source Location

• Repository: Apache_Shardingsphere
• File: KillProcessHandler.java
• Lines: 1-75

Signature

public final class KillProcessHandler implements GlobalDataChangedEventHandler {

    public NodePath getSubscribedNodePath()

    public Collection<Type> getSubscribedTypes()

    public void handle(final ContextManager contextManager, final DataChangedEvent event)
}

Import

import org.apache.shardingsphere.mode.manager.cluster.dispatch.handler.global.node.process.KillProcessHandler;

I/O Contract

Inputs

Name	Type	Required	Description
contextManager	ContextManager	Yes	Provides access to instance context and persist services
event	DataChangedEvent	Yes	Contains the event key (path), value, and type (ADDED/DELETED)

Outputs

Name	Type	Description
void	void	Side effects: kills SQL process on matching instance, cleans trigger path, or notifies lock

Usage Examples

// This handler is auto-loaded via SPI. Trigger a kill by persisting to the trigger path:
String triggerPath = NodePathGenerator.toPath(
    new KillProcessTriggerNodePath(new InstanceProcessNodeValue(instanceId, processId)));
repository.persist(triggerPath, "");
// The KillProcessHandler on the target instance will pick up the ADDED event and kill the process.

Related Pages

• Principle:Apache_Shardingsphere_Cluster_Metadata_Synchronization