Ansible Role: Prometheus

Install and configure Prometheus.

Requirements

An Ansible 2.2 or higher installation.
This role makes use of the Ansible json_filter that requires jmespath to be installed on the Ansible machine. See the requirements.txt file for further details on the specific version of jmespath required by the role.

Role Variables

Available variables are listed below, along with default values (see defaults/main.yml):

prometheus_release_tag: "latest"

The Prometheus release to be installed. By default, the latest release published at https://github.com/prometheus/prometheus/releases.

prometheus_release_url: ""

If set, the role will download Prometheus from the provided URL instead of using the download URL indicated in the Prometheus Github release metadata.

prometheus_user: "prometheus"
prometheus_group: "prometheus"

Prometheus system user and group.

prometheus_install_path: "/opt"

Directory containing the downloaded Prometheus release artifacts.

prometheus_bin_path: "/usr/local/bin"

Directory to which the Prometheus binaries will be symlinked.

prometheus_config_path: "/etc/prometheus"
prometheus_config_file: "prometheus.yml"

Prometheus configuration file and directory

prometheus_config: {}

YAML dictionary holding the Prometheus configuration. The complete Prometheus configuration reference can be found at https://prometheus.io/docs/prometheus/latest/configuration/configuration/.
NOTE: the provided prometheus configuration will be merged with the default one defined in vars/main.yml.

prometheus_rule_files_path: "{{ prometheus_config_path }}/rules"
prometheus_rule_files: {}
# e.g.
# prometheus_rule_files:
# alertmanager: |
#   groups:
#     - name: alertmanager.rules
#       rules:
#         - alert: AlertmanagerNotificationsFailing
#           expr: rate(alertmanager_notifications_failed_total[1m]) > 0
#           for: 1m
#           labels:
#             service: alertmanager
#             severity: critical
#           annotations:
#             title: Alertmanager is failing sending notications
#             description: Alertmanager is seeing errors for integration {{$labels.integration}}

Dictionary of Prometheus rule files to be created into the specified prometheus_rule_files_path.

prometheus_listen_address: "127.0.0.1:9090"

The Prometheus WebServer listen ip address and port.
NOTE: the Prometheus metrics will be available at {{ prometheus_listen_address }}/metrics.

prometheus_tsdb_path: "/var/lib/prometheus"
prometheus_tsdb_retention: |-
  {%- if prometheus_release_tag == 'latest' or prometheus_release_tag | regex_replace('^v(.*)$', '\\1') is version_compare('2.0.0', '>=') -%}
  15d
  {%- else -%}
  360h0m0s
  {%- endif -%}

Directory containing the Prometheus time-series database files. By default, the old data will be deleted after 14 days. An in depth analysis of the Operational Aspects of Prometheus Storage is available at https://prometheus.io/docs/prometheus/latest/storage/#operational-aspects.

prometheus_log_level: "info"

Prometheus deamon log verbosity level.

prometheus_additional_cli_args: ""

Additional command-line arguments to be added to the Prometheus service unit. For the complete refence of the available CLI arguments please refer to the output of the prometheus --help command.

Dependencies

None.

Example Playbooks

$ cat playbook.yml
- name: "Install and configure Prometheus"
  hosts: all
  roles:
    - { role: atosatto.prometheus }

Testing

Tests are automated with Molecule.

$ pip install tox

To test all the scenarios run

$ tox

To run a custom molecule command

$ tox -e py27-ansible29 -- molecule test -s prometheus-latest

License

MIT

Author Information

Andrea Tosatto (@_hilbert_)