java - JMS 性能和 RemoteTCPConnection.receive

Question

并行 JMS 消费者线程后，我没有获得所需的性能提升。
在 jvisualvm 中，我可以看到以下 jms 内容成为 self time CPU 列中的“获胜者”。

com.ibm.mq.jmqi.remote.impl.RemoteTCPConnection.receive()

这是线程实际执行的完整堆栈跟踪。

"RcvThread: com.ibm.mq.jmqi.remote.impl.RemoteTCPConnection@9167029[qmid=MWTEST53_2019-06-10_12.32.59,fap=12,channel=MA                  ,ccsid=1208,sharecnv=10,hbint=300,peer=MWTEST/162.11.56.86(1453),localport=61965,ssl=no,hConns=0,LastDataSend=1583310588163 (16ms ago),LastDataRecv=1583310588163 (16ms ago),]" - Thread t@651
   java.lang.Thread.State: RUNNABLE
        at java.net.SocketInputStream.socketRead0(Native Method)
        at java.net.SocketInputStream.socketRead(Unknown Source)
        at java.net.SocketInputStream.read(Unknown Source)
        at java.net.SocketInputStream.read(Unknown Source)
        at com.ibm.mq.jmqi.remote.impl.RemoteTCPConnection.receive(RemoteTCPConnection.java:1535)
        at com.ibm.mq.jmqi.remote.impl.RemoteRcvThread.receiveBuffer(RemoteRcvThread.java:794)
        at com.ibm.mq.jmqi.remote.impl.RemoteRcvThread.receiveOneTSH(RemoteRcvThread.java:757)
        at com.ibm.mq.jmqi.remote.impl.RemoteRcvThread.run(RemoteRcvThread.java:150)
        at java.lang.Thread.run(Unknown Source)

   Locked ownable synchronizers:
        - None

"jms_reader_thread_9" - Thread t@63
   java.lang.Thread.State: TIMED_WAITING
        at java.lang.Object.wait(Native Method)
        - waiting on <711f08> (a com.ibm.mq.jmqi.remote.impl.RemoteProxyQueue$GetterEventMonitor)
        at com.ibm.mq.jmqi.remote.impl.RemoteProxyQueue.proxyMQGET(RemoteProxyQueue.java:2492)
        at com.ibm.mq.jmqi.remote.api.RemoteFAP.jmqiGetInternalWithRecon(RemoteFAP.java:6536)
        at com.ibm.mq.jmqi.remote.api.RemoteFAP.jmqiGetInternal(RemoteFAP.java:6432)
        at com.ibm.mq.jmqi.internal.JmqiTools.getMessage(JmqiTools.java:1259)
        at com.ibm.mq.jmqi.remote.api.RemoteFAP.jmqiGet(RemoteFAP.java:6395)
        at com.ibm.mq.ese.jmqi.InterceptedJmqiImpl.jmqiGet(InterceptedJmqiImpl.java:910)
        at com.ibm.mq.ese.jmqi.ESEJMQI.jmqiGet(ESEJMQI.java:362)
        at com.ibm.mq.MQDestination.internalGetMessage(MQDestination.java:1012)
        - locked <1a7d0fa> (a com.ibm.mq.MQQueue)
        at com.ibm.mq.MQDestination.getInt(MQDestination.java:585)
        - locked <1a7d0fa> (a com.ibm.mq.MQQueue)
        at com.ibm.mq.MQDestination.get(MQDestination.java:456)
        at com.company.app.connection.MqConnection.nextRecordObject(MqConnection.java:351)
        at com.company.app.source.AppMqReader.next(AppMqReader.java:37)
        at com.company.app.source.AppJMSReadConnector.delegateNext(AppJMSReadConnector.java:202)
        at com.company.app.source.LegacyReadConnector.readNextWithRetry(LegacyReadConnector.java:114)
        at com.company.app.source.LegacyReadConnector.processNext(LegacyReadConnector.java:68)
        at com.company.app.source.AppJMSReadConnector.processNext(AppJMSReadConnector.java:435)
        at com.company.app.source.AppReadConnector.next(AppReadConnector.java:131)
        at com.company.app.source.AppReadConnector.next(AppReadConnector.java:114)
        at com.company.app.configuration.jaxb.ReadPipeline.run(ReadPipeline.java:262)
        at com.company.app.configuration.jaxb.AppAdaptor.run(AppAdaptor.java:684)
        at com.company.app.configuration.jaxb.RouteService.runOneIteration(RouteService.java:78)
        at com.google.common.util.concurrent.AbstractScheduledService$ServiceDelegate$Task.run(AbstractScheduledService.java:195)
        at com.google.common.util.concurrent.AbstractScheduledService$CustomScheduler$ReschedulableCallable.call(AbstractScheduledService.java:482)
        at com.google.common.util.concurrent.AbstractScheduledService$CustomScheduler$ReschedulableCallable.call(AbstractScheduledService.java:448)
        at com.google.common.util.concurrent.Callables$3.call(Callables.java:89)
        at java.util.concurrent.FutureTask.run(Unknown Source)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(Unknown Source)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(Unknown Source)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
        at java.lang.Thread.run(Unknown Source)

   Locked ownable synchronizers:
        - locked <cc8421> (a java.util.concurrent.locks.ReentrantLock$NonfairSync)

        - locked <15571ee> (a java.util.concurrent.ThreadPoolExecutor$Worker)

线程正在等待!几乎 10 个线程都处于 TIMED_WAITING 状态。这是 receive() 方法逻辑的代码片段

synchronized (this.getterEventMonitor) {
    if ((this.status & 0x4) == 0) {
        if (waitInterval > 0) {
            this.getterEventMonitor.wait(waitInterval);
        }
        else {

            this.getterEventMonitor.wait();
        }

    }
} 

状态 0x4 应为 ST_GETTER_SIGNALLED。

问题:等待 ST_GETTER_SIGNALLED 状态背后的逻辑是什么，线程实际在做什么以及如何使其更快？

Answer 1

这不仅仅是一条评论，而且，恕我直言，也比一个答案少一点，但因为对于评论来说，这里的文字太多了。

您确实需要描述线程何时生成并指定工作要做，因为很难找出可能出现的问题，但根据您的陈述，线程大多处于等待状态，那么.. .

...可能发生的情况是，如果所有 10 个线程共享相同的同步逻辑，那么在任何给定点，其中 9 个线程将被锁定在该逻辑之外。第 10 次，将等到您对 (this.status & 0x4) == 0) 的检查成功。

即。如果您的第 10 个线程处于此逻辑中并正在等待，则所有其他 9 个线程将等待第 10 个线程停止等待。

我认为您需要重新设计逻辑，以拥有一个监视事件的调度程序线程，然后将工作单元分派(dispatch)给等待线程。