Storm不追上最大的花销

Question

我已经创建了一个示例拓扑来测试设置max spout Expensing属性。这是一个简单的拓扑，有一个喷嘴和一个螺栓。喷口发出100000个元组，而螺栓在睡眠一秒钟后发出嘎嘎声。我已将“最大喷口支出”属性设置为10。我假设这意味着，如果一个喷口的未确认消息计数为10，那么该喷口将不会发出任何元组。但当我运行拓扑时，我可以看到喷口发出2160条消息，然后等待。我的理解是正确的还是遗漏了什么。我使用的是storm 0.9.5。下面是代码

public static void main(String[] args) {

    TopologyBuilder builder = new TopologyBuilder();
    builder.setSpout("spout", new TestSpout(), 1);
    builder.setBolt("bolt", new TestBolt(),1).shuffleGrouping("spout");
    Config conf = new Config();
    conf.setNumWorkers(1);
    conf.setMaxSpoutPending(10);
    try {
        StormSubmitter.submitTopology("test", conf, builder.createTopology());
    } catch (AlreadyAliveException e) {
        e.printStackTrace();
    } catch (InvalidTopologyException e) {
        e.printStackTrace();
    }
}


public class TestSpout extends BaseRichSpout {
private SpoutOutputCollector collector;
private int count = 1;

@Override
public void declareOutputFields(OutputFieldsDeclarer declarer) {
    declarer.declare(new Fields("spoutData"));
}

@Override
public void open(Map conf, TopologyContext context, SpoutOutputCollector collector) {
    this.collector = collector;
    System.out.println(context.maxTopologyMessageTimeout());
}

@Override
public void nextTuple() {

    if(count <= 100000) {
        System.out.println("Emitting : " + count);
        collector.emit(new Values(count++ + ""));
    }
}

}

public class TestBolt extends BaseRichBolt {
private OutputCollector collector;

@Override
public void prepare(Map stormConf, TopologyContext context, OutputCollector collector) {
    this.collector = collector;
}

@Override
public void execute(Tuple input) {
    try {
        System.out.println(input.getString(0));
        Thread.sleep(1000);
        collector.ack(input);
    } catch (InterruptedException e) {
        e.printStackTrace();
        System.out.println("Exception");
    }
}

@Override
public void declareOutputFields(OutputFieldsDeclarer declarer) {

}

}

Answer 1

您需要为您在Spout.nextTuple（）方法中发出的元组分配消息ID。否则，参数max.spout.pending将被忽略。例如，您可以使用您的count变量作为ID（基本上，任何东西都可以用作ID。它必须是唯一的。）

@Override
public void nextTuple() {
    if(count <= 100000) {
        System.out.println("Emitting : " + count);
        collector.emit(new Values(count++ + ""), count);
    }
}

否则，Storm无法将输出元组链接到bolt中确认的元组，也就是说，Storm无法计算有多少元组挂起。Storm只能跟踪ID为的元组。