spring cloud 集成 prometheus

2019独角兽企业重金招聘Python工程师标准>>>

由于spring boot的metrics与prometheus的metrics在格式上并不兼容.前者是json格式输出,后者是text格式输出. 
在使用prometheus对服务或者方法进行监控的时候, 需要对spring boot metrics进行转换,或者自己增加prometheus的metrics信息. 本文通过创建自定义注解的形式,引入prometheus, 只需要在被监控的方法外启用注解即可.

引入依赖

<dependency>
    <groupId>io.prometheus</groupId>
    <artifactId>simpleclient_spring_boot</artifactId>
    <version>0.0.26</version>
</dependency>

• 1
• 2
• 3
• 4
• 5

启动类增加注解

@SpringBootApplication
@EnablePrometheusEndpoint
@EnableSpringBootMetricsCollector
public class Application extends SpringBootServletInitializer {

@Override
protected ApplicationBuilder configure(
SpringApplicationBuilder application) {
return application.sources(Application.class);
} 

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10

自定义注解

import java.lang.annotation.*;

@Target(ElementType.METHOD)
@Retention(RetentionPolicy.RUNTIME)
@Documented
public @interface PrometheusMetrics {
    /**
     *  默认为空,程序使用method signature作为Metric name
     *  如果name有设置值,使用name作为Metric name
     * @return
     */
    String name() default "";

}

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14

创建Aspect , 对注解方法增加Metrics记录

@Aspect
@Component
@Slf4j
public class PrometheusMetricsAspect {
    private static final Counter requestTotal = Counter.build().name("couter_all").labelNames("api").help
            ("total request couter of api").register();
    private static final Counter requestError = Counter.build().name("couter_error").labelNames("api").help
            ("response Error couter of api").register();
    private static final Histogram histogram = Histogram.build().name("histogram_consuming").labelNames("api").help
            ("response consuming of api").register();

    @Pointcut("@annotation(com.ubbor.common.aspect.PrometheusMetrics)")
    public void pcMethod() {
    }

    @Around(value="pcMethod() && @annotation(annotation)")
    public Object MetricsCollector(ProceedingJoinPoint joinPoint, PrometheusMetrics annotation) throws Throwable {

        MethodSignature methodSignature = (MethodSignature) joinPoint.getSignature();
        PrometheusMetrics prometheusMetrics = methodSignature.getMethod().getAnnotation(PrometheusMetrics.class);
        if (prometheusMetrics != null) {
            String name;
            if (StringUtil.isNotEmpty(prometheusMetrics.name()) ){
                name = prometheusMetrics.name();
            }else{
                HttpServletRequest request = ((ServletRequestAttributes) RequestContextHolder.getRequestAttributes())
                        .getRequest();
                name = request.getRequestURI();
            }

            requestTotal.labels(name).inc();
            Histogram.Timer requestTimer = histogram.labels(name).startTimer();
            Object object;
            try {
                object = joinPoint.proceed();
            } catch (Exception e) {
                requestError.labels(name).inc();
                throw e;
            } finally {
                requestTimer.observeDuration();
            }

            return object;
        } else {
            return joinPoint.proceed();
        }

    }
}

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41
• 42
• 43
• 44
• 45
• 46
• 47
• 48
• 49

给需要监控的方法增加自定义注解

@PrometheusMetrics
@RequestMapping(value = "/test", method = RequestMethod.POST)
@ResponseBody
public ResponseMessage  execute  (RequestMessage request) throws Exception {
//do something
}

• 1
• 2
• 3
• 4
• 5
• 6

增加配置

spring.metrics.servo.enabled=false

• 1

去掉重复的metrics，不然在prometheus的控制台的targets页签里，会一直显示此endpoint为down状态

检查输出

http://localhost:9001/prometheus

参考资料 
https://prometheus.io/docs/introduction/getting_started 
https://github.com/prometheus/docs/blob/master/content/docs/operating/configuration.md

转载于:https://my.oschina.net/xiaominmin/blog/1788585