Prometheus exposes metrics about itself under the /metrics HTTP path, just like most other instrumented applications would. You can see an example of such a metrics endpoint at https://demo.promlabs.com/metrics.

There are too many metrics to list here, but they fall into the following prefixed groups:

• go_*: Standard metrics about the Go runtime (heap metrics, garbage collection statistics, etc.).
• process_*: Standard metrics about the Prometheus process from the operating system's point of view (memory usage, CPU usage, file descriptor usage, etc.).
• net_conntrack_dialer_conn_*: TCP connection metrics for each scrape pool.
• prometheus_*: All metrics exposed explicitly by Prometheus itself. There are many sub-categories here:
  • prometheus_api_*: HTTP API metrics.
  • prometheus_build_info: An info metric containing Go and git build information labels about the Prometheus server binary.
  • prometheus_config_*: Metrics about configuration loading and reloading states.
  • prometheus_engine_*: Metrics about the PromQL query engine.
  • prometheus_http_*: Generic statistics about HTTP requests to both the web API and the web UI.
  • prometheus_notifications_*: Metrics about the notification manager that sends firing alerts to the Alertmanager.
  • prometheus_remote_storage_*: Metrics about interactions with remote storage endpoints.
  • prometheus_rule_*: Rule (group) execution statistics.
  • prometheus_sd_*: Metrics related to service discovery activity.
  • prometheus_target_*: Metrics related to target scraping and metadata handling.
  • prometheus_template_*: Statistics about text template expansions and errors.
  • prometheus_treecache_*: Zookeeper-treecache-related metrics that are only relevant if you are using Zookeeper-based service discovery.
  • prometheus_tsdb_*: Prometheus TSDB metrics about ingested samples, checkpoints, chunk sizes, loaded blocks, errors, and more.
  • prometheus_web_federation_*: Federation-related metrics.
  • promhttp_*: HTTP metrics about the /metrics handler itself.

You can study all metrics in detail by reading the HELP comment strings in your Prometheus server's /metrics endpoint for each metric.

The Alertmanager exposes metrics and can be monitored in just the same way as Prometheus.