JAVA Metrics 度量工具入门

最新推荐文章于 2022-04-20 17:19:48 发布

it噩梦

最新推荐文章于 2022-04-20 17:19:48 发布

阅读量672

点赞数 1

分类专栏： jvm 文章标签： jvm

本文链接：https://blog.csdn.net/qq_37362891/article/details/121631138

版权

jvm 专栏收录该内容

24 篇文章 0 订阅

订阅专栏

在这里插入图片描述

文章目录

前言

Metrics是一个给JAVA服务的各项指标提供度量工具的包，在JAVA代码中嵌入Metrics代码，可以方便的对业务代码的各个指标进行监控。

官方文档：

github:https://github.com/dropwizard/metrics
官方文档：https://metrics.dropwizard.io/4.2.0/getting-started.html

环境搭建

        <dependency>
            <groupId>io.dropwizard.metrics</groupId>
            <artifactId>metrics-core</artifactId>
            <version>3.2.6</version>
        </dependency>

五大metric

Meter

主要用来测量服务接口的调用频率。

public class MeterExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private final static Meter requestMeter = registry.meter("tqs");

    private final static Meter sizeMeter = registry.meter("volume");

    public static void main(String[] args) {
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
                .convertRatesTo(TimeUnit.MINUTES)
                .convertDurationsTo(TimeUnit.MINUTES).build();
        reporter.start(10, TimeUnit.SECONDS);
        for (; ; ) {
            upload(new byte[current().nextInt(1000)]);
            randomSleep();
        }
    }

    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(10));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

    private static void upload(byte[] bytes) {
        requestMeter.mark();
        sizeMeter.mark(bytes.length);
    }
}

-- Meters ----------------------------------------------------------------------
tqs
             count = 2
         mean rate = 11.80 events/minute
     1-minute rate = 12.00 events/minute
     5-minute rate = 12.00 events/minute
    15-minute rate = 12.00 events/minute
volume
             count = 900
         mean rate = 5310.19 events/minute
     1-minute rate = 4714.49 events/minute
     5-minute rate = 4610.97 events/minute
    15-minute rate = 4593.04 events/minute


21-12-2 16:14:34 ===============================================================

-- Meters ----------------------------------------------------------------------
tqs
             count = 3
         mean rate = 8.93 events/minute
     1-minute rate = 11.12 events/minute
     5-minute rate = 11.80 events/minute
    15-minute rate = 11.93 events/minute
volume
             count = 1824
         mean rate = 5428.88 events/minute
     1-minute rate = 4877.28 events/minute
     5-minute rate = 4643.08 events/minute
    15-minute rate = 4603.72 events/minute

可以看到这些数据一分钟平均速率8.93，每分钟上传的字节数是5428.88字节。

Guage

返回一个时刻的value值。

SimpleGauge

public class SimpleGaugeExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();

    private static Queue<Long> queue = new LinkedBlockingDeque<>();


    public static void main(String[] args) {
        registry.register(MetricRegistry.name(SimpleGaugeExample.class), (Gauge<Integer>)queue::size);
        reporter.start(1,TimeUnit.SECONDS);
        new Thread(() -> {
            for (; ; ) {
                randomSleep();
                queue.add(System.nanoTime());
            }
        }).start();

        new Thread(() -> {
            for (; ; ) {
                randomSleep();
                queue.poll();
            }
        }).start();
    }

    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(5));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}

运行上面代码，可以看到，每隔10秒，对queue的size进行输出。

-- Gauges ----------------------------------------------------------------------
fast.cloud.nacos.metrics.SimpleGaugeExample
             value = 11


21-12-2 16:19:05 ===============================================================

-- Gauges ----------------------------------------------------------------------
fast.cloud.nacos.metrics.SimpleGaugeExample
             value = 11


21-12-2 16:19:06 ===============================================================

-- Gauges ----------------------------------------------------------------------
fast.cloud.nacos.metrics.SimpleGaugeExample
             value = 11

我们可以看到生产者、消费者的速率，以及队列的挤压情况。

JmxAttributeGauge

public class JmxAttributeGaugeExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();

    public static void main(String[] args) throws MalformedObjectNameException, InterruptedException {
        reporter.start(10,TimeUnit.SECONDS);
        registry.register(MetricRegistry.name(JmxAttributeGaugeExample.class, "HeapMemory"), new JmxAttributeGauge(new ObjectName("java.lang:type=Memory"), "HeapMemoryUsage"));
        registry.register(MetricRegistry.name(JmxAttributeGaugeExample.class, "NonHeapMemory"), new JmxAttributeGauge(new ObjectName("java.lang:type=Memory"), "NonHeapMemoryUsage"));
        Thread.currentThread().join();
    }
}

-- Gauges ----------------------------------------------------------------------
fast.cloud.nacos.metrics.metric.JmxAttributeGaugeExample.HeapMemory
             value = javax.management.openmbean.CompositeDataSupport(compositeType=javax.management.openmbean.CompositeType(name=java.lang.management.MemoryUsage,items=((itemName=committed,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=init,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=max,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=used,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)))),contents={committed=257425408, init=268435456, max=3817865216, used=16163152})
fast.cloud.nacos.metrics.metric.JmxAttributeGaugeExample.NonHeapMemory
             value = javax.management.openmbean.CompositeDataSupport(compositeType=javax.management.openmbean.CompositeType(name=java.lang.management.MemoryUsage,items=((itemName=committed,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=init,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=max,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=used,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)))),contents={committed=10944512, init=2555904, max=-1, used=10009504})


21-12-2 16:21:47 ===============================================================

-- Gauges ----------------------------------------------------------------------
fast.cloud.nacos.metrics.metric.JmxAttributeGaugeExample.HeapMemory
             value = javax.management.openmbean.CompositeDataSupport(compositeType=javax.management.openmbean.CompositeType(name=java.lang.management.MemoryUsage,items=((itemName=committed,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=init,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=max,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=used,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)))),contents={committed=257425408, init=268435456, max=3817865216, used=16163152})
fast.cloud.nacos.metrics.metric.JmxAttributeGaugeExample.NonHeapMemory
             value = javax.management.openmbean.CompositeDataSupport(compositeType=javax.management.openmbean.CompositeType(name=java.lang.management.MemoryUsage,items=((itemName=committed,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=init,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=max,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)),(itemName=used,itemType=javax.management.openmbean.SimpleType(name=java.lang.Long)))),contents={committed=10944512, init=2555904, max=-1, used=10130232})

可以看到，每个十秒输出JVM堆内存及非堆内存的信息。

RadioGauge

用于创建两个数字之间的某种比率，比如业务受理的成功率或者失败率等。

public class RadioGauge {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();

    private final static Meter totalMeter = new Meter();

    private final static Meter successMeter = new Meter();

    public static void main(String[] args) {
        reporter.start(10,TimeUnit.SECONDS);

        registry.gauge("success-rate", () -> new RatioGauge() {
            @Override
            protected Ratio getRatio() {
                return Ratio.of(successMeter.getCount(), totalMeter.getCount());
            }
        });
        for (; ; ) {
            shortSleep();
            business();
        }
    }

    private static void business() {
        totalMeter.mark();
        try {
            int x = 10 / current().nextInt(6);
            successMeter.mark();
        } catch (Exception e) {
            System.out.println("ERROR");
        }
    }

    private static void shortSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(6));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}

21-12-2 16:32:18 ===============================================================

-- Gauges ----------------------------------------------------------------------
success-rate
             value = 1.0


ERROR
21-12-2 16:32:28 ===============================================================

-- Gauges ----------------------------------------------------------------------
success-rate
             value = 0.75


21-12-2 16:32:38 ===============================================================

-- Gauges ----------------------------------------------------------------------
success-rate
             value = 0.875

每隔十秒，业务的成功率度量信息将会输出到控制台上。

CacheGauge

我们对Gauge的实时性要求没有那么高，就可以使用cached gauge。

public class CacheGauge {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();


    public static void main(String[] args) throws InterruptedException {
        reporter.start(10,TimeUnit.SECONDS);

        registry.gauge("cache-db-size", () -> new CachedGauge(30,TimeUnit.SECONDS) {
            @Override
            protected Object loadValue() {
                return queryFromDbSize();
            }
        });

        Thread.currentThread().join();
    }

    private static Object queryFromDbSize() {
        System.out.println("=====queryFromDbSize====");
        return System.currentTimeMillis();
    }
}


21-12-2 16:34:08 ===============================================================

-- Gauges ----------------------------------------------------------------------
cache-db-size
=====queryFromDbSize====
             value = 1638434048893


21-12-2 16:34:18 ===============================================================

-- Gauges ----------------------------------------------------------------------
cache-db-size
             value = 1638434048893


21-12-2 16:34:28 ===============================================================

-- Gauges ----------------------------------------------------------------------
cache-db-size
             value = 1638434048893


21-12-2 16:34:38 ===============================================================

-- Gauges ----------------------------------------------------------------------
cache-db-size
             value = 1638434048893

可以看到没有输出=queryFromDbSize的时候，这时候就是取得缓存中的数据。

Counter

我们每次对query取size是有性能影响的，Counter metric 使用64的递增或者递减，避免了性能的侵入。

public class CounterExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();
    private static final BlockingQueue<Long> queue = new LinkedBlockingQueue<>();

    public static void main(String[] args) throws InterruptedException {
        reporter.start(10, TimeUnit.SECONDS);
        Counter counter = registry.counter("queue-count", Counter::new);
        new Thread(() -> {
            for (; ; ) {
                randomSleep();
                queue.add(System.nanoTime());
                counter.inc();
            }
        });
        new Thread(() -> {
            for (; ; ) {
                randomSleep();
                if (queue.poll() != null) {
                    counter.dec();
                }
            }
        });

        Thread.currentThread().join();
    }


    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(500));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}

-- Counters --------------------------------------------------------------------
queue-count
             count = 0


21-12-2 16:40:38 ===============================================================

-- Counters --------------------------------------------------------------------
queue-count
             count = 0


21-12-2 16:40:48 ===============================================================

-- Counters --------------------------------------------------------------------
queue-count
             count = 0

Histogram

是一种统计报告图

public class HistogramExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();
    private static final Histogram histogram = registry.histogram("search-result");

    public static void main(String[] args) {
        reporter.start(10, TimeUnit.SECONDS);

        while (true) {
            doSearch();
            randomSleep();
        }
    }

    private static void doSearch() {
        histogram.update(current().nextInt(10));
    }

    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(10));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

}

-- Histograms ------------------------------------------------------------------
search-result
             count = 3
               min = 0
               max = 5
              mean = 1.95
            stddev = 2.16
            median = 1.00
              75% <= 5.00
              95% <= 5.00
              98% <= 5.00
              99% <= 5.00
            99.9% <= 5.00


21-12-2 16:42:59 ===============================================================

-- Histograms ------------------------------------------------------------------
search-result
             count = 4
               min = 0
               max = 9
              mean = 3.89
            stddev = 3.65
            median = 5.00
              75% <= 9.00
              95% <= 9.00
              98% <= 9.00
              99% <= 9.00
            99.9% <= 9.00

可以看到75%的搜索结构都小于9。

count: 参与统计的数据有多少条
min: 统计数据的最小值
max: 统计数据的最大值
mean: 统计数据的中间值
stddev: 标准方差
median: 统计数据的中间值
75%:

Timers

Timer其实是 Histogram 和 Meter 的结合， histogram 某部分代码/调用的耗时， meter统计TPS。

public class TimerTest {

    public static Random random = new Random();

    public static void main(String[] args) throws InterruptedException {
        MetricRegistry registry = new MetricRegistry();
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry).build();
        reporter.start(1, TimeUnit.SECONDS);

        Timer timer = registry.timer(MetricRegistry.name(TimerTest.class,"get-latency"));

        Timer.Context ctx;

        while(true){
            ctx = timer.time();
            Thread.sleep(random.nextInt(1000));
            ctx.stop();
        }

    }

}

21-12-2 16:49:06 ===============================================================

-- Timers ----------------------------------------------------------------------
fast.cloud.nacos.metrics.metric.TimerTest.get-latency
             count = 7
         mean rate = 1.75 calls/second
     1-minute rate = 0.00 calls/second
     5-minute rate = 0.00 calls/second
    15-minute rate = 0.00 calls/second
               min = 125.79 milliseconds
               max = 746.72 milliseconds
              mean = 474.94 milliseconds
            stddev = 217.70 milliseconds
            median = 385.33 milliseconds
              75% <= 704.84 milliseconds
              95% <= 746.72 milliseconds
              98% <= 746.72 milliseconds
              99% <= 746.72 milliseconds
            99.9% <= 746.72 milliseconds


21-12-2 16:49:07 ===============================================================

-- Timers ----------------------------------------------------------------------
fast.cloud.nacos.metrics.metric.TimerTest.get-latency
             count = 10
         mean rate = 2.00 calls/second
     1-minute rate = 0.00 calls/second
     5-minute rate = 0.00 calls/second
    15-minute rate = 0.00 calls/second
               min = 125.79 milliseconds
               max = 971.09 milliseconds
              mean = 493.35 milliseconds
            stddev = 251.51 milliseconds
            median = 368.69 milliseconds
              75% <= 704.84 milliseconds
              95% <= 971.09 milliseconds
              98% <= 971.09 milliseconds
              99% <= 971.09 milliseconds
            99.9% <= 971.09 milliseconds

四种reporter

ConsoleReporter

public class ConsoleReporterExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();

    private static final Histogram histogram = registry.histogram("search-result");

    public static void main(String[] args) {
        reporter.start(10,TimeUnit.SECONDS);

        while (true) {
            doSearch();
            randomSleep();
        }
    }

    private static void doSearch() {
        histogram.update(current().nextInt(10));
    }


    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(500));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

}

顾名思义，输出到控制台:

-- Histograms ------------------------------------------------------------------
search-result
             count = 1
               min = 9
               max = 9
              mean = 9.00
            stddev = 0.00
            median = 9.00
              75% <= 9.00
              95% <= 9.00
              98% <= 9.00
              99% <= 9.00
            99.9% <= 9.00

JmxReporter

通过jconsole查看

public class JmxReporterExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final JmxReporter reporter = JmxReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();
    private static final Histogram histogram = registry.histogram("search-result");

    public static void main(String[] args) {
        reporter.start();

        while (true) {
            doSearch();
            randomSleep();
        }
    }

    private static void doSearch() {
        histogram.update(current().nextInt(10));
    }


    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(500));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

}

在这里插入图片描述

CsvReporter

public class CsvReporterExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final CsvReporter reporter = CsvReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build(new File("/Users/qinfuxiang/Downloads"));
    private static final Histogram histogram = registry.histogram("search-result");

    public static void main(String[] args) {
        reporter.start(10,TimeUnit.SECONDS);

        while (true) {
            doSearch();
            randomSleep();
        }
    }

    private static void doSearch() {
        histogram.update(current().nextInt(10));
    }


    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(500));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

}

会在指定目录下导出数据到csv中。
在这里插入图片描述

Slf4jReporter

通过log输出到控制台

public class SLF4jReporterExample {
    private static final MetricRegistry registry = new MetricRegistry();

    private static final Slf4jReporter reporter = Slf4jReporter.forRegistry(registry)
            .outputTo(LoggerFactory.getLogger(SLF4jReporterExample.class))
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.SECONDS).build();

    private static final Histogram histogram = registry.histogram("search-result");

    public static void main(String[] args) {
        reporter.start(10,TimeUnit.SECONDS);

        while (true) {
            doSearch();
            randomSleep();
        }
    }

    private static void doSearch() {
        histogram.update(current().nextInt(10));
    }


    private static void randomSleep() {
        try {
            TimeUnit.SECONDS.sleep(current().nextInt(500));
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

}

17:07:17.866 [metrics-logger-reporter-1-thread-1] INFO fast.cloud.nacos.metrics.reporter.SLF4jReporterExample - type=HISTOGRAM, name=search-result, count=1, min=3, max=3, mean=3.0, stddev=0.0, median=3.0, p75=3.0, p95=3.0, p98=3.0, p99=3.0, p999=3.0
17:07:27.852 [metrics-logger-reporter-1-thread-1] INFO fast.cloud.nacos.metrics.reporter.SLF4jReporterExample - type=HISTOGRAM, name=search-result, count=1, min=3, max=3, mean=3.0, stddev=0.0, median=3.0, p75=3.0, p95=3.0, p98=3.0, p99=3.0, p999=3.0
17:07:37.853 [metrics-logger-reporter-1-thread-1] INFO fast.cloud.nacos.metrics.reporter.SLF4jReporterExample - type=HISTOGRAM, name=search-result, count=1, min=3, max=3, mean=3.0, stddev=0.0, median=3.0, p75=3.0, p95=3.0, p98=3.0, p99=3.0, p999=3.0
17:07:47.851 [metrics-logger-reporter-1-thread-1] INFO fast.cloud.nacos.metrics.reporter.SLF4jReporterExample - type=HISTOGRAM, name=search-result, count=1, min=3, max=3, mean=3.0, stddev=0.0, median=3.0, p75=3.0, p95=3.0, p98=3.0, p99=3.0, p999=3.0

it噩梦

关注

1
点赞
踩
0

收藏

觉得还不错? 一键收藏
0
评论
JAVA Metrics 度量工具入门

前言Metrics是一个给JAVA服务的各项指标提供度量工具的包，在JAVA代码中嵌入Metrics代码，可以方便的对业务代码的各个指标进行监控。官方文档：github:https://github.com/dropwizard/metrics官方文档：https://metrics.dropwizard.io/4.2.0/getting-started.html环境搭建 <dependency> <groupId>io.dro.
复制链接

扫一扫