项目运行指标：micrometer自定义metrics

micrometer自定义metrics

micrometer提供了基于Java的monitor facade，其与springboot应用和prometheus的集成方式如下图展示

上图中展示的很清楚，应用通过micrometer采集和暴露监控端点给prometheus，prometheus通过pull模式来采集监控时序数据信息。之后作为数据源提供给grafana进行展示。

micrometer支持的度量方式及在springboot中的应用示例

Counter
Counter(计数器)简单理解就是一种只增不减的计数器。它通常用于记录服务的请求数量、完成的任务数量、错误的发生数量等等。

1. package com.dxz.producter.monitor;
2.  
3. import org.springframework.stereotype.Service;
4.  
5. import io.micrometer.core.instrument.Counter;
6. import io.micrometer.core.instrument.Metrics;
7.  
8. @Service("collectorService")
9. public class CollectorService {
10.  
11. static final Counter userCounter = Metrics.counter("user.counter.total", "services", "demo");
12.  
13. public void processCollectResult() throws InterruptedException {
14.  
15. while (true) {
16. userCounter.increment(1D);
17. }
18. }
19. }

Gauge
Gauge(仪表)是一个表示单个数值的度量，它可以表示任意地上下移动的数值测量。Gauge通常用于变动的测量值，如当前的内存使用情况，同时也可以测量上下移动的"计数"，比如队列中的消息数量。

1. package com.dxz.producter.monitor;
2.  
3. import java.util.ArrayList;
4. import java.util.List;
5. import java.util.concurrent.atomic.AtomicInteger;
6.  
7. import org.springframework.stereotype.Component;
8.  
9. import io.micrometer.core.instrument.Gauge;
10. import io.micrometer.core.instrument.ImmutableTag;
11. import io.micrometer.core.instrument.Metrics;
12. import io.micrometer.core.instrument.Tag;
13. import io.micrometer.core.instrument.simple.SimpleMeterRegistry;
14.  
15. @Component("passCaseMetric")
16. public class PassCaseMetric {
17.  
18. List<Tag> init() {
19. ArrayList<Tag> list = new ArrayList() {
20. };
21. list.add(new ImmutableTag("service", "demo"));
22. return list;
23. }
24.  
25. AtomicInteger atomicInteger = new AtomicInteger(0);
26.  
27. Gauge passCaseGuage = Gauge.builder("pass.cases.guage", atomicInteger, AtomicInteger::get).tag("service", "demo")
28. .description("pass cases guage of demo").register(new SimpleMeterRegistry());
29.  
30. AtomicInteger passCases = Metrics.gauge("pass.cases.guage.value", init(), atomicInteger);
31.  
32. public void handleMetrics() {
33.  
34. while (true) {
35. if (System.currentTimeMillis() % 2 == 0) {
36. passCases.addAndGet(100);
37. System.out.println("ADD + " + passCaseGuage.measure() + " : " + passCases);
38. } else {
39. int val = passCases.addAndGet(-100);
40. if (val < 0) {
41. passCases.set(1);
42. }
43. System.out.println("DECR - " + passCaseGuage.measure() + " : " + passCases);
44. }
45. }
46.  
47. }
48.  
49. }

增加一个controller，触发他们：

1. package com.dxz.producter.web;
2.  
3. import org.springframework.beans.factory.annotation.Autowired;
4. import org.springframework.web.bind.annotation.RequestMapping;
5. import org.springframework.web.bind.annotation.RequestMethod;
6. import org.springframework.web.bind.annotation.RestController;
7.  
8. import com.dxz.producter.monitor.CollectorService;
9. import com.dxz.producter.monitor.PassCaseMetric;
10.  
11. @RestController
12. @RequestMapping("/monitor")
13. public class MonitorController {
14.  
15. @Autowired
16. CollectorService collectorService;
17.  
18. @Autowired
19. PassCaseMetric passCaseMetric;
20.  
21. @RequestMapping(value = "/counter", method = RequestMethod.GET)
22. public String counter() throws InterruptedException {
23. collectorService.processCollectResult();
24. return "+1";
25. }
26.  
27. @RequestMapping(value = "/gauge", method = RequestMethod.GET)
28. public String gauge() throws InterruptedException {
29. passCaseMetric.handleMetrics();
30. return "+gauge";
31. }
32.  
33. }

启动springboot应用，可以在http://host:port/actuator/prometheus 看到端点收集到的数据。其他的也是类似的不再一一截图展示。

这里使用了一个true的循环用来展示不断更新的效果。

 

同样的可以在grafana中看到监控展示信息

 

Timer
Timer(计时器)同时测量一个特定的代码逻辑块的调用(执行)速度和它的时间分布。简单来说，就是在调用结束的时间点记录整个调用块执行的总时间，适用于测量短时间执行的事件的耗时分布，例如消息队列消息的消费速率。

 

1. @Test
2. public void testTimerSample(){
3. Timer timer = Timer.builder("timer")
4. .tag("timer", "timersample")
5. .description("timer sample test.")
6. .register(new SimpleMeterRegistry());
7.  
8. for(int i=0; i<2; i++) {
9. timer.record(() -> {
10. try {
11. TimeUnit.SECONDS.sleep(2);
12. }catch (InterruptedException e){
13.  
14. }
15.  
16. });
17. }
18.  
19. System.out.println(timer.count());
20. System.out.println(timer.measure());
21. System.out.println(timer.totalTime(TimeUnit.SECONDS));
22. System.out.println(timer.mean(TimeUnit.SECONDS));
23. System.out.println(timer.max(TimeUnit.SECONDS));
24. }

响应数据

2
[Measurement{statistic='COUNT', value=2.0}, Measurement{statistic='TOTAL_TIME', value=4.005095763}, Measurement{statistic='MAX', value=2.004500494}]
4.005095763
2.0025478815
2.004500494

Summary
Summary(摘要)用于跟踪事件的分布。它类似于一个计时器，但更一般的情况是，它的大小并不一定是一段时间的测量值。在micrometer中，对应的类是DistributionSummary，它的用法有点像Timer，但是记录的值是需要直接指定，而不是通过测量一个任务的执行时间。

 

1. @Test
2. public void testSummary(){
3.  
4. DistributionSummary summary = DistributionSummary.builder("summary")
5. .tag("summary", "summarySample")
6. .description("summary sample test")
7. .register(new SimpleMeterRegistry());
8.  
9. summary.record(2D);
10. summary.record(3D);
11. summary.record(4D);
12.  
13. System.out.println(summary.count());
14. System.out.println(summary.measure());
15. System.out.println(summary.max());
16. System.out.println(summary.mean());
17. System.out.println(summary.totalAmount());
18. }

 

响应数据：

3
[Measurement{statistic='COUNT', value=3.0}, Measurement{statistic='TOTAL', value=9.0}, Measurement{statistic='MAX', value=4.0}]
4.0
3.0
9.0

 

序

本文主要研究下如何使用自定义micrometer的metrics

实例

DemoMetrics

1. public class DemoMetrics implements MeterBinder {
2. AtomicInteger count = new AtomicInteger(0);
3.  
4. @Override
5. public void bindTo(MeterRegistry meterRegistry) {
6. Gauge.builder("demo.count", count, c -> c.incrementAndGet())
7. .tags("host", "localhost")
8. .description("demo of custom meter binder")
9. .register(meterRegistry);
10. }
11. }

这里实现了MeterBinder接口的bindTo方法，将要采集的指标注册到MeterRegistry

注册

• 原始方式

new DemoMetrics().bindTo(registry);

• springboot autoconfigure

1. @Bean
2. public DemoMetrics demoMetrics(){
3. return new DemoMetrics();
4. }

在springboot只要标注下bean，注入到spring容器后，springboot会自动注册到registry。springboot已经帮你初始化了包括UptimeMetrics等一系列metrics。详见源码解析部分。

验证

curl -i http://localhost:8080/actuator/metrics/demo.count

返回实例

1. {
2. "name": "demo.count",
3. "measurements": [
4. {
5. "statistic": "VALUE",
6. "value": 6
7. }
8. ],
9. "availableTags": [
10. {
11. "tag": "host",
12. "values": [
13. "localhost"
14. ]
15. }
16. ]
17. }

源码解析

MetricsAutoConfiguration

spring-boot-actuator-autoconfigure-2.0.0.RELEASE-sources.jar!/org/springframework/boot/actuate/autoconfigure/metrics/MetricsAutoConfiguration.java

1. @Configuration
2. @ConditionalOnClass(Timed.class)
3. @EnableConfigurationProperties(MetricsProperties.class)
4. @AutoConfigureBefore(CompositeMeterRegistryAutoConfiguration.class)
5. public class MetricsAutoConfiguration {
6.  
7. @Bean
8. @ConditionalOnMissingBean
9. public Clock micrometerClock() {
10. return Clock.SYSTEM;
11. }
12.  
13. @Bean
14. public static MeterRegistryPostProcessor meterRegistryPostProcessor(
15. ApplicationContext context) {
16. return new MeterRegistryPostProcessor(context);
17. }
18.  
19. @Bean
20. @Order(0)
21. public PropertiesMeterFilter propertiesMeterFilter(MetricsProperties properties) {
22. return new PropertiesMeterFilter(properties);
23. }
24.  
25. @Configuration
26. @ConditionalOnProperty(value = "management.metrics.binders.jvm.enabled", matchIfMissing = true)
27. static class JvmMeterBindersConfiguration {
28.  
29. @Bean
30. @ConditionalOnMissingBean
31. public JvmGcMetrics jvmGcMetrics() {
32. return new JvmGcMetrics();
33. }
34.  
35. @Bean
36. @ConditionalOnMissingBean
37. public JvmMemoryMetrics jvmMemoryMetrics() {
38. return new JvmMemoryMetrics();
39. }
40.  
41. @Bean
42. @ConditionalOnMissingBean
43. public JvmThreadMetrics jvmThreadMetrics() {
44. return new JvmThreadMetrics();
45. }
46.  
47. @Bean
48. @ConditionalOnMissingBean
49. public ClassLoaderMetrics classLoaderMetrics() {
50. return new ClassLoaderMetrics();
51. }
52.  
53. }
54.  
55. @Configuration
56. static class MeterBindersConfiguration {
57.  
58. @Bean
59. @ConditionalOnClass(name = { "ch.qos.logback.classic.LoggerContext",
60. "org.slf4j.LoggerFactory" })
61. @Conditional(LogbackLoggingCondition.class)
62. @ConditionalOnMissingBean(LogbackMetrics.class)
63. @ConditionalOnProperty(value = "management.metrics.binders.logback.enabled", matchIfMissing = true)
64. public LogbackMetrics logbackMetrics() {
65. return new LogbackMetrics();
66. }
67.  
68. @Bean
69. @ConditionalOnProperty(value = "management.metrics.binders.uptime.enabled", matchIfMissing = true)
70. @ConditionalOnMissingBean
71. public UptimeMetrics uptimeMetrics() {
72. return new UptimeMetrics();
73. }
74.  
75. @Bean
76. @ConditionalOnProperty(value = "management.metrics.binders.processor.enabled", matchIfMissing = true)
77. @ConditionalOnMissingBean
78. public ProcessorMetrics processorMetrics() {
79. return new ProcessorMetrics();
80. }
81.  
82. @Bean
83. @ConditionalOnProperty(name = "management.metrics.binders.files.enabled", matchIfMissing = true)
84. @ConditionalOnMissingBean
85. public FileDescriptorMetrics fileDescriptorMetrics() {
86. return new FileDescriptorMetrics();
87. }
88.  
89. }
90.  
91. static class LogbackLoggingCondition extends SpringBootCondition {
92.  
93. @Override
94. public ConditionOutcome getMatchOutcome(ConditionContext context,
95. AnnotatedTypeMetadata metadata) {
96. ILoggerFactory loggerFactory = LoggerFactory.getILoggerFactory();
97. ConditionMessage.Builder message = ConditionMessage
98. .forCondition("LogbackLoggingCondition");
99. if (loggerFactory instanceof LoggerContext) {
100. return ConditionOutcome.match(
101. message.because("ILoggerFactory is a Logback LoggerContext"));
102. }
103. return ConditionOutcome
104. .noMatch(message.because("ILoggerFactory is an instance of "
105. + loggerFactory.getClass().getCanonicalName()));
106. }
107.  
108. }
109.  
110. }

可以看到这里注册了好多metrics，比如UptimeMetrics，JvmGcMetrics，ProcessorMetrics，FileDescriptorMetrics等

这里重点看使用@Bean标注了MeterRegistryPostProcessor

MeterRegistryPostProcessor

spring-boot-actuator-autoconfigure-2.0.0.RELEASE-sources.jar!/org/springframework/boot/actuate/autoconfigure/metrics/MeterRegistryPostProcessor.java

1. class MeterRegistryPostProcessor implements BeanPostProcessor {
2.  
3. private final ApplicationContext context;
4.  
5. private volatile MeterRegistryConfigurer configurer;
6.  
7. MeterRegistryPostProcessor(ApplicationContext context) {
8. this.context = context;
9. }
10.  
11. @Override
12. public Object postProcessAfterInitialization(Object bean, String beanName)
13. throws BeansException {
14. if (bean instanceof MeterRegistry) {
15. getConfigurer().configure((MeterRegistry) bean);
16. }
17. return bean;
18. }
19.  
20. @SuppressWarnings("unchecked")
21. private MeterRegistryConfigurer getConfigurer() {
22. if (this.configurer == null) {
23. this.configurer = new MeterRegistryConfigurer(beansOfType(MeterBinder.class),
24. beansOfType(MeterFilter.class),
25. (Collection<MeterRegistryCustomizer<?>>) (Object) beansOfType(
26. MeterRegistryCustomizer.class),
27. this.context.getBean(MetricsProperties.class).isUseGlobalRegistry());
28. }
29. return this.configurer;
30. }
31.  
32. private <T> Collection<T> beansOfType(Class<T> type) {
33. return this.context.getBeansOfType(type).values();
34. }
35.  
36. }

可以看到这里new了一个MeterRegistryConfigurer，重点注意这里使用beansOfType(MeterBinder.class)方法的返回值给其构造器

MeterRegistryConfigurer

spring-boot-actuator-autoconfigure-2.0.0.RELEASE-sources.jar!/org/springframework/boot/actuate/autoconfigure/metrics/MeterRegistryConfigurer.java

1. class MeterRegistryConfigurer {
2.  
3. private final Collection<MeterRegistryCustomizer<?>> customizers;
4.  
5. private final Collection<MeterFilter> filters;
6.  
7. private final Collection<MeterBinder> binders;
8.  
9. private final boolean addToGlobalRegistry;
10.  
11. MeterRegistryConfigurer(Collection<MeterBinder> binders,
12. Collection<MeterFilter> filters,
13. Collection<MeterRegistryCustomizer<?>> customizers,
14. boolean addToGlobalRegistry) {
15. this.binders = (binders != null ? binders : Collections.emptyList());
16. this.filters = (filters != null ? filters : Collections.emptyList());
17. this.customizers = (customizers != null ? customizers : Collections.emptyList());
18. this.addToGlobalRegistry = addToGlobalRegistry;
19. }
20.  
21. void configure(MeterRegistry registry) {
22. if (registry instanceof CompositeMeterRegistry) {
23. return;
24. }
25. // Customizers must be applied before binders, as they may add custom
26. // tags or alter timer or summary configuration.
27. customize(registry);
28. addFilters(registry);
29. addBinders(registry);
30. if (this.addToGlobalRegistry && registry != Metrics.globalRegistry) {
31. Metrics.addRegistry(registry);
32. }
33. }
34.  
35. @SuppressWarnings("unchecked")
36. private void customize(MeterRegistry registry) {
37. LambdaSafe.callbacks(MeterRegistryCustomizer.class, this.customizers, registry)
38. .withLogger(MeterRegistryConfigurer.class)
39. .invoke((customizer) -> customizer.customize(registry));
40. }
41.  
42. private void addFilters(MeterRegistry registry) {
43. this.filters.forEach(registry.config()::meterFilter);
44. }
45.  
46. private void addBinders(MeterRegistry registry) {
47. this.binders.forEach((binder) -> binder.bindTo(registry));
48. }
49.  
50. }

可以看到configure方法里头调用了addBinders，也就是把托管给spring容器的MeterBinder实例bindTo到meterRegistry

小结

springboot2引入的micrometer，自定义metrics只需要实现MeterBinder接口，然后托管给spring即可，springboot的autoconfigure帮你自动注册到meterRegistry。

 

项目运行指标：micrometer自定义metrics_meterregistrycustomizer-CSDN博客

SpringBoot结合prometheus自定义埋点实践_springboot prometheus 自定义-CSDN博客