Quarkus BackPressure配置

问题描述

我使用quarkus smallrye反应性消息和kafka获得了以下堆栈跟踪：

2020-07-24 01:38:31,662 ERROR [io.sma.rea.mes.kafka] (executor-thread-870) SRMSG18207: Unable to dispatch message to Kafka: io.smallrye.mutiny.subscription.BackPressureFailure: Could not emit tick 211 due to lack of requests
        at io.smallrye.mutiny.operators.multi.builders.IntervalMulti$IntervalRunnable.run(IntervalMulti.java:83)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
        at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)
        at org.jboss.threads.ContextClassLoaderSavingRunnable.run(ContextClassLoaderSavingRunnable.java:35)
        at org.jboss.threads.EnhancedQueueExecutor.safeRun(EnhancedQueueExecutor.java:2046)
        at org.jboss.threads.EnhancedQueueExecutor$ThreadBody.doRunTask(EnhancedQueueExecutor.java:1578)
        at org.jboss.threads.EnhancedQueueExecutor$ThreadBody.run(EnhancedQueueExecutor.java:1452)
        at org.jboss.threads.DelegatingRunnable.run(DelegatingRunnable.java:29)
        at org.jboss.threads.ThreadLocalResettingRunnable.run(ThreadLocalResettingRunnable.java:29)
        at java.lang.Thread.run(Thread.java:748)
        at org.jboss.threads.JBossThread.run(JBossThread.java:479)

所以我读了https://smallrye.io/smallrye-mutiny/#_how_do_i_control_the_back_pressure

根据文档，我添加了BackPressure控件。

之前：

@Outgoing( "eqs-crossing-xxx" )
    public Multi< EQSAlert > eqsCrossingXXX_XXX(){

        final String series = CrossingEnum.XXX_XXX.getSeries();
        final String equipment = CrossingEnum.XXX_XXX.getEquipment();

        final String vehicleRegex = vehicleRegexService.getRegexBySeries( series );

        log.info( "Incoming request for {} - {}",series,equipment);
        log.info( "Vehicle regex : {}",vehicleRegex );

        return Multi
                .createFrom()
                .ticks()
                .every(
                        Duration.ofSeconds( poolingInterval )
                )
                .concatMap(i -> {
                    final Multi<CrossingState> crossingStateBySeriesAndEquipment = CrossingState.getCrossingStateBySeriesAndEquipment(client,equipment);

                    return crossingStateBySeriesAndEquipment.flatMap(crossingState ->
                            crossingState.isActive() ?
                                    EQSAlert.getEQSAlertBySeriesAndEquipment(
                                            client,vehicleRegex,equipment
                                    )
                                    :
                                    Multi.createFrom().empty()
                    );
                });
    }

之后：

@Outgoing( "eqs-crossing-xxx" )
    public Multi< EQSAlert > eqsCrossingXXX_XXX(){

        final String series = CrossingEnum.XXX_XXX.getSeries();
        final String equipment = CrossingEnum.XXX_XXX.getEquipment();

        final String vehicleRegex = vehicleRegexService.getRegexBySeries( series );

        log.info( "Incoming request for {} - {}",vehicleRegex );

        return Multi
                .createFrom()
                .ticks()
                .every(
                        Duration.ofSeconds( poolingInterval )
                )
                .onOverflow()
                .buffer(10)
                .concatMap(i -> {
                    final Multi<CrossingState> crossingStateBySeriesAndEquipment = CrossingState.getCrossingStateBySeriesAndEquipment(client,equipment
                                    )
                                    :
                                    Multi.createFrom().empty()
                    );
                });
    }

现在一切都好了。

我的帖子的目的是了解我为什么需要这样做？

https://smallrye.io/smallrye-mutiny/apidocs/io/smallrye/mutiny/subscription/BackPressureStrategy.html

为什么缓冲区不能保留u？

如您所见，我每5秒钟执行一次简单的sql函数调用（poolingInterval）。该函数返回一些记录（通过池返回不超过10条记录）

所以流量很低

请让我说些什么来了解缓冲区管理。

谢谢

解决方法

您的下游单位时间内只能消耗一定数量的物品。这取决于您在做什么。默认情况下，Kafka将自身限制为5个并发写入（您可以配置它）。

因此，如果排放更多，则以无界的方式（不是背压感知的）方式，下游无法跟上。添加10个项目的缓冲区可能会解决一些小问题，但这可能还不够。

drop更为激进；如果下游无法跟上，则仅丢弃这些物品。 dropPreviousItems正在删除已收到的项目。