Overview

  • grafana-agent 的 metrics采集，完全兼容 prometheus exporter 生态，一些常见的 exporter，会在 grafana-agent 中内嵌实现（列表如下）;
  • 对于未嵌入到 grafana-agent中的 exporter，则可以在 grafana-agent 中配置 scrape_configs 来完成抓取和收集，请参考抓取第三方exporter;

grafana-agent 内置实现的 exporter 列表

内置exporter的配置项说明

grafana-agent 本身的配置

  1. server:
  2.   log_level: info
  3.   http_listen_port: 12345

grafana-agent 抓取 metrics 的相关配置（类似于prometheus的scrape_configs）

  1. metrics:
  2.   global:
  3.     scrape_interval: 15s
  4.     scrape_timeout: 10s
  5.     remote_write:
  6.       - url: https://n9e-server:19000/prometheus/v1/write
  7.         basic_auth:
  8.           username: <string>
  9.           password: <string>
  11. # grafana-agent integration 相关的配置
  12. integrations:
  13.   ## grafana-agent self-integration
  14.   ## grafana-agent 本身的metrics 采集，这也是一个内嵌的 integration，可以选择启用或者关闭。
  15.   agent:
  16.     ### 是否开启针对grafana-agent 自身的integration，允许grafana-agent自动采集和发送其自身的metrics
  17.     [enabled: <boolean> | default = false]
  19.     # Sets an explicit value for the instance label when the integration is
  20.     # self-scraped. Overrides inferred values.
  21.     #
  22.     # The default value for this integration is inferred from the agent hostname
  23.     # and HTTP listen port, delimited by a colon.
  24.     [instance: <string>]
  26.     # Automatically collect metrics from this integration. If disabled,
  27.     # the agent integration will be run but not scraped and thus not
  28.     # remote_written. Metrics for the integration will be exposed at
  29.     # /integrations/agent/metrics and can be scraped by an external process.
  30.     ### 这个配置项如果设置为false，那么 /integrations/agent/metrics 的数据并不会被自动抓取和发送
  31.     ### 但是，该接口 /integrations/agent/metrics 的数据仍然支持被外部的抓取进程所抓取 
  32.     [scrape_integration: <boolean> | default = <integrations_config.scrape_integrations>]
  34.     # How often should the metrics be collected? Defaults to
  35.     # prometheus.global.scrape_interval.
  36.     [scrape_interval: <duration> | default = <global_config.scrape_interval>]
  38.     # The timeout before considering the scrape a failure. Defaults to
  39.     # prometheus.global.scrape_timeout.
  40.     [scrape_timeout: <duration> | default = <global_config.scrape_timeout>]
  42.     # How frequent to truncate the WAL for this integration.
  43.     [wal_truncate_frequency: <duration> | default = "60m"]
  45.     # Allows for relabeling labels on the target.
  46.     relabel_configs:
  47.       [- <relabel_config> ... ]
  49.     # Relabel metrics coming from the integration, allowing to drop series
  50.     # from the integration that you don't care about.
  51.     metric_relabel_configs:
  52.       [ - <relabel_config> ... ]
  54.   # Client TLS Configuration
  55.   # Client Cert/Key Values need to be defined if the server is requesting a certificate
  56.   #  (Client Auth Type = RequireAndVerifyClientCert || RequireAnyClientCert).
  57.   http_tls_config: <tls_config>
  59.   ## 控制内嵌的 node_exporter 工作逻辑 
  60.   node_exporter: <node_exporter_config>
  62.   ## 控制内嵌的 process_exporter 工作逻辑
  63.   process_exporter: <process_exporter_config>
  65.   ## 控制内嵌的 mysqld_exporter 工作逻辑
  66.   mysqld_exporter: <mysqld_exporter_config>
  68.   ## 控制内嵌的 redis_exporter 工作逻辑
  69.   redis_exporter: <redis_exporter_config>
  71.   ## 控制内嵌的 dnsmasq_exporter 工作逻辑
  72.   dnsmasq_exporter: <dnsmasq_exporter_config>
  74.   ## 控制内嵌的 elasticsearch_exporter 工作逻辑
  75.   elasticsearch_expoter: <elasticsearch_expoter_config>
  77.   # Controls the memcached_exporter integration
  78.   memcached_exporter: <memcached_exporter_config>
  80.   ## 控制内嵌的 postgres_exporter 工作逻辑
  81.   postgres_exporter: <postgres_exporter_config>
  83.   ## 控制内嵌的 statsd_exporter 工作逻辑
  84.   statsd_exporter: <statsd_exporter_config>
  86.   ## 控制内嵌的 consul_exporter 工作逻辑
  87.   consul_exporter: <consul_exporter_config>
  89.   ## 控制内嵌的 windows_exporter 工作逻辑
  90.   windows_exporter: <windows_exporter_config>
  92.   ## 控制内嵌的 kafka_exporter 工作逻辑
  93.   kafka_exporter: <kafka_exporter_config>
  95.   ## 控制内嵌的 mongodb_exporter 工作逻辑 
  96.   mongodb_exporter: <mongodb_exporter_config>
  98.   ## 控制内嵌的 github_exporter 工作逻辑
  99.   github_exporter: <github_exporter_config>
  101.   # Automatically collect metrics from enabled integrations. If disabled,
  102.   # integrations will be run but not scraped and thus not remote_written. Metrics
  103.   # for integrations will be exposed at /integrations/<integration_key>/metrics
  104.   # and can be scraped by an external process.
  105.   ## 如果设置为false，相关的exporter metrics接口仍会被暴露出来，但是grafana-agent不会去主动抓取和发送
  106.   [scrape_integrations: <boolean> | default = true]
  108.   # Extra labels to add to all samples coming from integrations.
  109.   labels:
  110.     { <string>: <string> }
  112.   # The period to wait before restarting an integration that exits with an
  113.   # error.
  114.   [integration_restart_backoff: <duration> | default = "5s"]
  116.   # A list of remote_write targets. Defaults to global_config.remote_write.
  117.   # If provided, overrides the global defaults.
  118.   prometheus_remote_write:
  119.     - [<remote_write>]

通过grafana-agent抓取第三方exporter并收集

如文章开头所述，对于未嵌入到grafana-agent中的exporter，则可以在grafana-agent中配置scrape_configs来完成抓取和收集，其配置形式完全等同于 prometheus scrape_configs。

grafana-agent中关于自定义配置scrape_configs的详细说明如下：

  1. # scrape_configs like prometheus style
  2. configs:
  3.   scrape_timeout: 10s
  5.   # 比如，我们可以配置抓取 grafana-agent 本身的 metrics ： http://127.0.0.1:12345/metrics
  6.   - name: grafana-agent
  7.     host_filter: false
  8.     scrape_configs:
  9.       - job_name: grafana-agent
  10.         static_configs:
  11.           - targets: ['127.0.0.1:12345']
  12.     remote_write:
  13.       - url: http://localhost:9090/api/v1/write
  15.   # 再比如,我们也可以配置抓取您的应用程序暴露的metrics接口： http://helloworld.app:8088/metrics
  16.   - name: outside-exporters
  17.     host_filter: false
  18.     scrape_configs:
  19.       - job_name: prometheus
  20.         static_configs:
  21.           - targets: ['127.0.0.1:9090']
  22.             labels:
  23.               cluster: 'fc-monitoring'
  24.     remote_write:
  25.       - url: https://n9e-server:19000/prometheus/v1/write
  26.         basic_auth:
  27.           username: <string>
  28.           password: <string>