java结合prometheus如何实现自定义数据监控
作者：涅米涅米
文章介绍了如何配置Prometheus监控系统,包括配置文件prometheus.yml、被监控应用的指标暴露配置以及自定义监控指标的实现,同时,还详细说明了监控应用如何通过Prometheus API获取数据、处理数据并返回结果
一、配置prometheus
- prometheus.yml
... - job_name: 'my-service' metrics_path: /metrics static_configs: - targets: ['xxx.xxx.xxx.xxx:yyyy'] //被监控应用的url ...
二、被监控应用
思路
- 引入相关依赖
- 配置监控指标暴露的endpoint
- 自定义监控指标
关键代码
1. 引入相关依赖
- pom.xml
<!-- prometheus --> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient</artifactId> <version>0.3.0</version> </dependency> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient_hotspot</artifactId> <version>0.3.0</version> </dependency> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient_servlet</artifactId> <version>0.3.0</version> </dependency>
2. 将监控指标暴露到’/metrics’
- PrometheusConfig.java
import io.prometheus.client.exporter.MetricsServlet;
import io.prometheus.client.hotspot.DefaultExports;
import org.springframework.boot.web.servlet.ServletRegistrationBean;
...
@Configuration
public class PrometheusConfig {
// ...
@Bean
public ServletRegistrationBean servletRegistrationBean(){
DefaultExports.initialize();
return new ServletRegistrationBean(new MetricsServlet(), "/metrics");
}
}3. 自定义监控指标
- MyMetrics.java
import io.prometheus.client.Counter;
import io.prometheus.client.Gauge;
...
// counter只增不减
static final Counter customizeCounter = Counter.build()
.name("customize_counter") //定义名称，名称可用于搜索
.help("customize counter") //定义描述
.register();
customizeCounter.inc(); //当前对象加1
// gauge可增可减
static final Gauge customizeGauge = Gauge.build()
.name("customize_gauge")
.help("customize gauge")
.labelNames("label1", "label2", "label3") //定义标签名称，可定义多个
.register();
customizeGauge.inc(); //当前对象加1
customizeGauge.dec(); //当前对象减1
customizeGauge.labels("value1","value2","value3").set(1100); //设置标签值，标签可用于条件筛选
// 此外还有histogram和summary两种指标三、监控应用
思路
- 引入相关依赖
- 调用prometheus API获取数据
- 整理数据并返回
关键代码
1. 引入相关依赖
- pom.xml
<!-- prometheus --> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient</artifactId> <version>0.3.0</version> </dependency> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient_hotspot</artifactId> <version>0.3.0</version> </dependency> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient_servlet</artifactId> <version>0.3.0</version> </dependency>
2. 调用prometheus API
- 获取某个时间点的数据
String query = "customize_counter"; String url = "http://[ip]:[port]/api/v1/query?query=" + query + "&time=2019-05-01T20:10:51.781Z"; HttpGet get = new HttpGet(url); CloseableHttpClient httpClient = HttpClients.custom().build(); CloseableHttpResponse response = httpClient.execute(get);
- 获取某个时间段的数据
String query = "rate(customize_counter{label1 = value1}[30s])";
String url = "http://[ip]:[port]/api/v1/query_range?query=" + query + "&start=2019-05-01T20:10:51.781Z&end=2019-05-02T20:10:51.781Z";
HttpGet get = new HttpGet(url);
CloseableHttpClient httpClient = HttpClients.custom().build();
CloseableHttpResponse response = httpClient.execute(get);更多使用方法请参考Prometheus - 查询
3. 处理数据
- prometheus有时会返回乱序的结果，以下代码可以按时间戳排序
public class ComparatorPromData implements Comparator {
@Override
public int compare(Object o1, Object o2) {
List list1 = (List)o1;
List list2 = (List)o2;
return ((Integer) list1.get(0)).compareTo(((Integer) list2.get(0)));
}
}public class SortUtil {
public static List sortPromData(List<List> list) {
ComparatorPromData comparator = new ComparatorPromData();
Collections.sort(list, comparator);
return list;
}
}总结
以上为个人经验，希望能给大家一个参考，也希望大家多多支持脚本之家。