4.1 KiB
Prometheus
Prometheus is a metrics collection and alerting monitoring solution.
Configuration
To enable this service, add the following configuration to your vars.yml file and re-run the installation process:
########################################################################
# #
# prometheus #
# #
########################################################################
prometheus_enabled: true
########################################################################
# #
# /prometheus #
# #
########################################################################
By default, Prometheus is configured to scrape (collect metrics from) its own process. If you wish to disable this behavior, use prometheus_self_process_scraper_enabled: false.
To make Prometheus useful, you'll need to make it scrape one or more hosts by adjusting the configuration.
Integrating with Prometheus Node Exporter
If you've installed Prometheus Node Exporter on the same host, you can make Prometheus scrape its metrics with the following additional configuration:
prometheus_self_node_scraper_enabled: true
prometheus_self_node_scraper_static_configs_target: "{{ prometheus_node_exporter_identifier }}:9100"
# node-exporter runs in another container network, so we need to connect to it.
prometheus_container_additional_networks:
- "{{ prometheus_node_exporter_container_network }}"
To scrape a remote Prometheus Node Exporter instance, do not use prometheus_self_node_scraper_*, but rather follow the Scraping any other exporter service guide below.
Scraping any other exporter service
To inject your own scrape configuration, use the prometheus_config_scrape_configs_additional variable that's part of the ansible-role-prometheus Ansible role.
Example additional configuration:
prometheus_config_scrape_configs_additional:
- job_name: some_job
metrics_path: /metrics
scrape_interval: 120s
scrape_timeout: 120s
static_configs:
- targets:
- some-host:8080
- job_name: another_job
metrics_path: /metrics
scrape_interval: 120s
scrape_timeout: 120s
static_configs:
- targets:
- another-host:8080
If you're scraping others services running in containers over the container network, make sure the Prometheus container is connected to their own network by adjusting prometheus_container_additional_networks as demonstrated above for Integrating with Prometheus Node Exporter.
Exposing the web interface
By setting a hostname you will expose prometheus on this domain. Usually you should also set up basic_auth in this case, otherwise everyone will be able to access your metrics
prometheus_hostname: prometheus.example.com
# Uncommenting the following lines allows you to configure basic auth
# prometheus_basic_auth_enabled: true
# prometheus_basic_auth_username: ''
# prometheus_basic_auth_password: ''
Recommended other services
- Grafana - a web-based tool for visualizing your Prometheus metrics (time-series)
- prometheus-blackbox-exporter - Blackbox probing of HTTP/HTTPS/DNS/TCP/ICMP and gRPC endpoints
- prometheus-node-exporter - an exporter for machine metrics
- prometheus-postgres-exporter - an exporter for monitoring a Postgres database server
- Healthchecks - a simple and Effective Cron Job Monitoring solution