ActiveMQ Artemis：CORE管理API与hawtio控制台性能问题求助

背景 #

• 使用版本：ActiveMQ Artemis 2.19.1
• 实现逻辑：通过CORE Management API（org.apache.activemq.artemis.api.core.client.*和org.apache.activemq.artemis.api.core.management.*）获取特定前缀队列的消费者数量，当消费者数为0时触发后端操作，该逻辑由20秒一次的定时任务轮询broker实现
• 当前问题：每次轮询请求都会创建临时地址，获取信息后再删除，导致Artemis/Hawtio Web控制台树状结构持续刷新，UI卡顿；日志中反复出现临时队列创建与销毁的记录

相关日志 #

[org.apache.activemq.artemis.core.server.plugin.impl] AMQ841007: created queue: QueueImpl[name=activemq.management.96a1005f-b67c-40af-b9cd-811f36750c5e, postOffice=PostOfficeImpl [server=ActiveMQServerImpl::name=artemis-2.19.1], temp=true]@58d1994e

[org.apache.activemq.artemis.core.server.plugin.impl] AMQ841008: destroyed queue: QueueImpl[name=activemq.management.96a1005f-b67c-40af-b9cd-811f36750c5e, postOffice=PostOfficeImpl [server=ActiveMQServerImpl::name=artemis-2.19.1], temp=true]@58d1994e, with args address: activemq.management.96a1005f-b67c-40af-b9cd-811f36750c5e, session: ServerSessionImpl(), checkConsumerCount: true, removeConsumers: false, autoDeleteAddress: true

当前代码实现 #

初始化代码 #

private ServerLocator locator;
private ClientSessionFactory defaultFactory;

public ManagementHelper(String defaultURL) {
    this.locator = ActiveMQClient.createServerLocator(defaultURL);
    this.defaultFactory = locator.createSessionFactory();
}

定时任务代码 #

@Scheduled(fixedRateString = "20000")
public void getConsumerNbos(ClientSessionFactory factory, ServerLocator locator) {
    try (ClientSession session = factory.createSession(this.username, this.password, false, true, true,
            locator.isPreAcknowledge(), locator.getAckBatchSize());
            ClientRequestor requestor = new ClientRequestor(session, "activemq.management");) {

        ClientMessage message = session.createMessage(false);
        ManagementHelper.putOperationInvocation(message, ResourceNames.BROKER, listAllConsumersAsJSON);
        session.start();

        ClientMessage replyConsumer = requestor.request(message);
        String resultJSON = (String) ManagementHelper.getResult(replyConsumer, String.class);

        ClientMessage message2 = session.createMessage(false);
        ManagementHelper.putOperationInvocation(message2, ResourceNames.BROKER, MANAGEMENT_OPERATION_QUEUES);

        ClientMessage replyQueueNames = requestor.request(message2);
        Object[] objQueueNames = (Object[]) ManagementHelper.getResult(replyQueueNames);
    } catch (Exception e) {

    }
}

疑问与需求 #

选择CORE API的核心原因是它能直接处理集群中的不同节点，相比Jolokia工作量更小。现寻求以下解决方案：

• 是否可以通过连接/会话池减少临时队列的创建频率？
• 能否配置broker在消费者数为0时主动通知后端，替代当前的定时轮询方式？
• 是否支持批量删除临时队列，而非逐个销毁？
• 其他可解决Hawtio控制台卡顿问题的方案？

内容的提问来源于stack exchange，提问作者Cat