prometheus - 如何在 Prometheus 中使用两个指标执行查询？

Question

我正在使用 Prometheus 从 Apache Flink 查询指标。我想测量 Map 函数每秒输入和输出的记录数。当我在 Prometheus 中查询两个不同的指标时，图表只显示其中之一。

flink_taskmanager_job_task_operator_numRecordsInPerSecond{operator_name="Map"} 
or flink_taskmanager_job_task_operator_numRecordsOutPerSecond{operator_name="Map"}

如果我改变运营商没有关系 or至 and .该图表仅显示第一个 ( flink_taskmanager_job_task_operator_numRecordsInPerSecond )。我也尝试过编辑 Prometheus 配置文件 /etc/prometheus/prometheus.yml但我对 Prometheus 没有太多经验，而且我的配置有问题。我的解决方案基于 this post .

global:
  scrape_interval: 15s

scrape_configs:
  - job_name: 'prometheus'
    scrape_interval: 5s
    static_configs:
      - targets: ['localhost:9090']
  - job_name: 'node_exporter'
    scrape_interval: 5s
    static_configs:
      - targets: ['localhost:9100']   
  - job_name: 'flink'
    scrape_interval: 5s
    static_configs:
      - targets: ['localhost:9250', 'localhost:9251', '192.168.56.20:9250']
    metrics_path: /
# HOW TO ADD THE OPERATOR NAME ON THE METRIC NAME?
    metric_relabel_configs:
      - source_labels: [__name__]
      regex: '(flink_taskmanager_job_task_operator)_(\w+)'
      replacement: '${2}'
      target_label: pool
      - source_labels: [__name__]
      regex: '(flink_taskmanager_job_task_operator)_(\w+)'
      replacement: '${1}_bytes'
      target_label: __name__

Answer 1

首先，对于更复杂的图形，您绝对应该调查 Grafana .
内置的 Prometheus 图很有用，例如。用于调试，但绝对更有限。特别是一张图只会显示一个查询的结果。

现在对于我绝对不推荐的 hack:

flink_taskmanager_job_task_operator_numRecordsInPerSecond{operator_name="Map"}
or
label_replace(flink_taskmanager_job_task_operator_numRecordsOutPerSecond{operator_name="Map"}, "distinct", "foo", "job", ".*")

从， as documented

vector1 or vector2 results in a vector that contains all original elements (label sets + values) of vector1 and additionally all elements of vector2 which do not have matching label sets in vector1.

您可以将第一个向量的标签中不存在的新标签添加到第二个向量，从而保留两者中的所有元素。