• 推送度量指标
    • Java批量任务例子

    推送度量指标

    偶尔你需要监控不能被获取的实例。它们可能被防火墙保护，或者它们生命周期太短而不能通过拉模式获取数据。Prometheus的Pushgateway允许你将这些实例的时间序列数据推送到Prometheus的代理任务中。结合Prometheus简单的文本导出格式，这使得即使没有客户库，也能使用shell脚本获取数据。

    • shell实现用例，查看Readme
    • Java, 详见PushGateway类
    • Go，详见Push和PushAdd
    • Python, 详见Pushgateway
    • Ruby, 详见Pushgateway

    Java批量任务例子

    这个例子主要说明, 如何执行一个批处理任务，并且在没有执行成功时报警

    如果使用Maven，添加下面的代码到pom.xml文件中：

    1.         <dependency>
    2.             <groupId>io.prometheus</groupId>
    3.             <artifactId>simpleclient</artifactId>
    4.             <version>0.0.10</version>
    5.         </dependency>
    6.         <dependency>
    7.             <groupId>io.prometheus</groupId>
    8.             <artifactId>simpleclient_pushgateway</artifactId>
    9.             <version>0.0.10</version>
    10.         </dependency>

    执行批量作业的代码：

    1. import io.prometheus.client.CollectorRegistry;
    2. import io.prometheus.client.Gauge;
    3. import io.prometheus.client.exporter.PushGateway;
    5. void executeBatchJob() throws Exception {
    6.  CollectorRegistry registry = new CollectorRegistry();
    7.  Gauge duration = Gauge.build()
    8.      .name("my_batch_job_duration_seconds")
    9.      .help("Duration of my batch job in seconds.")
    10.      .register(registry);
    11.  Gauge.Timer durationTimer = duration.startTimer();
    12.  try {
    13.    // Your code here.
    15.    // This is only added to the registry after success,
    16.    // so that a previous success in the Pushgateway is not overwritten on failure.
    17.    Gauge lastSuccess = Gauge.build()
    18.        .name("my_batch_job_last_success_unixtime")
    19.        .help("Last time my batch job succeeded, in unixtime.")
    20.        .register(registry);
    21.    lastSuccess.setToCurrentTime();
    22.  } finally {
    23.    durationTimer.setDuration();
    24.    PushGateway pg = new PushGateway("127.0.0.1:9091");
    25.    pg.pushAdd(registry, "my_batch_job");
    26.  }
    27. }

    警报一个Pushgateway，如果需要的话，修改host和port

    如果任务最近没有运行，请创建一个警报到Alertmanager。将以下内容添加到Pushgateway的Prometheus服务的记录规则中：record rules ALERT MyBatchJobNotCompleted IF min(time() - my_batch_job_last_success_unixtime{job="my_batch_job"}) > 60 * 60 FOR 5m WITH { severity="page" } SUMMARY "MyBatchJob has not completed successfully in over an hour"