Exporting and scraping Prometheus metrics

GitHub Enterprise Server includes a Prometheus metrics exporter to simplify observability and monitoring. This feature allows metrics to be scraped directly from the appliance, eliminating the need to forward collectd metrics to an external collectd server for transformation before sending them to a Prometheus system. By supporting Prometheus-compatible metrics natively, the exporter enables seamless integration with external monitoring dashboards and alerting systems.

Enabling Prometheus metrics exporter and scraping metrics

By default, the Prometheus metrics exporter and scraping are disabled. Enabling the exporter for collectd metrics opens the metrics endpoint (also knows as scrape endpoint) at http(s)://HOSTNAME:9103. Ensure that firewall rules allow traffic on port 9103.

You can enable the Prometheus metrics exporter using the 管理控制台, the command line interface within the appliance, or the API.

Enabling using the 管理控制台

在上的管理帐户中，在任一页面的右上角，单击。
如果你尚未在“站点管理员”页上，请在左上角单击“站点管理员”。
在“ 站点管理”边栏中，单击“管理控制台”。
In the monitoring settings, select Enable prometheus.
In the Trusted IPv4/IPv6 addresses or CIDR blocks in a comma-separated list field, type the comma-separated list of ip addresses or CIDR blocks. An example: 10.0.0.1, 192.168.1.0/8. If this field is left empty, then scraping requests from any source IP will be accepted.
在“设置”边栏下，单击“保存设置”。

Note

保存管理控制台中的设置会重启系统服务，这可能会导致用户可察觉的停机时间。
等待配置运行完毕。

Enabling via the command line interface

To enable the Prometheus metrics exporter, SSH into the GitHub Enterprise Server appliance and run the following commands:

ghe-config app.prometheus.enabled true
# Specify a comma-separated list of trusted IPv4/IPv6 addresses or CIDR blocks.
ghe-config app.prometheus.trusted-ips "10.0.0.1, 192.168.1.0/8"
ghe-config-apply

Enabling using the API

You can enable the Prometheus metrics exporter using the API by sending a PUT request to update the configuration settings.

curl -L -H "Content-Type: application/json" -X PUT -u "api_key:xxxxxxx" https://<hostname>/manage/v1/config/settings -d '{"prometheus": {"enabled": true, "trusted_ips": "10.0.0.1, 192.168.1.0/8"}}'

Verifying the Prometheus metrics exporter is enabled

To verify that the Prometheus metrics exporter is enabled, use curl to query the /metrics endpoint on port 9103. For more information about the administrative ports, see 网络端口.

curl localhost:9103/metrics

If successful, the response will include metrics with the collectd_ prefix.

Configuring your external Prometheus server

To scrape exported metrics from GitHub Enterprise Server, add a scraping configuration to the prometheus.yaml file on your external Prometheus server.

- job_name: "ghes-prom-agent"
  scrape_interval: 10s
  static_configs:
    - targets: ["http(s)://HOSTNAME:9103/metrics"]

Support for the darkshipped Prometheus metrics exporter

If you have been using the Prometheus metrics exporter since it was darkshipped in GitHub Enterprise Server 3.12, it will continue to work after upgrading to 3.16. In this version, you now have the option to specify trusted IPs for scraping requests.

You can share your feedback on the Prometheus metrics exporter and scraping functionality with GitHub. For help with questions or issues, visit GitHub Enterprise 支持.

Exporting and scraping Prometheus metrics

本文内容