BlockingWaitStrategy CPU 100%

[QinZhiWenXuan]

I used BlockingWaitStrategy in my code,it makes my project use CPU 100%

I use it for exception alert.Our system tps is not high,Fewer exceptions. As you know,I didn't use YieldingWaitStrategy,I used BlockingWaitStrategy. It performed so well for a long time, But today it used up my CPU.

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0"
         xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>
    <dependencies>
    <!-- Omit other dependencies here -->
    <!-- disruptor -->
    <dependency>
        <groupId>com.lmax</groupId>
        <artifactId>disruptor</artifactId>
        <version>4.0.0.RC1</version>
    </dependency>
    </dependencies>
</project>

    public void init() {
        this.disruptor = new Disruptor<>(
                new ReportPayloadFactory(),
                1024 * 4,
                new NamedThreadFactory("owl-disruptor-worker-", Boolean.TRUE),
                ProducerType.MULTI,
                new BlockingWaitStrategy()
        );
        this.disruptor
                .handleEventsWith(new ReportConsumer(this.messageService))
                .then(new ClearingReportPayloadEventHandler())
        ;
        this.disruptor.start();
        this.producer = new ReportProducer(this.disruptor.getRingBuffer());
    }

Desktop (please complete the following information):

• OS: [Linux]
• Version [4.0.0.RC1]
• JVM Version [openjdk version "11.0.6"]

[MendelMonteiro]

You should profile the CPU usage just to be sure where the time is being taken.

Having said that I suspect that if your first consumer is much slower than your second consumer the second blocking wait strategy will be spinning in this loop waiting for a dependent sequence: https://github.com/LMAX-Exchange/disruptor/blob/master/src/main/java/com/lmax/disruptor/BlockingWaitStrategy.java#L45

In the .NET port there is an implementation that will not use 100% CPU as it will actually sleep the thread occasionally instead of just yielding.

https://github.com/disruptor-net/Disruptor-net/blob/master/src/Disruptor/BlockingSpinWaitWaitStrategy.cs

[QinZhiWenXuan]

You should profile the CPU usage just to be sure where the time is being taken.

Having said that I suspect that if your first consumer is much slower than your second consumer the second blocking wait strategy will be spinning in this loop waiting for a dependent sequence: https://github.com/LMAX-Exchange/disruptor/blob/master/src/main/java/com/lmax/disruptor/BlockingWaitStrategy.java#L45

In the .NET port there is an implementation that will not use 100% CPU as it will actually sleep the thread occasionally instead of just yielding.

https://github.com/disruptor-net/Disruptor-net/blob/master/src/Disruptor/BlockingSpinWaitWaitStrategy.cs

Thank you for your reply.Maybe I didn't describe it clearly.My business scenario is a simple asynchronous alarm,Each consumer only delivers messages to MQ.There should be no chance of any one being blocked.Let me refer to the implementation.Finally thank you anyway