dmarcs-metrics-exporter

Export metrics derived from DMARC aggregate reports to Prometheus. This exporter regularly polls for new aggregate report emails via IMAP. The following metrics will be collected and exposed at an HTTP endpoint for Prometheus:

• dmarc_total: Total number of reported messages.
• dmarc_compliant_total: Total number of DMARC compliant messages.
• dmarc_quarantine_total: Total number of quarantined messages.
• dmarc_reject_total: Total number of rejected messages.
• dmarc_spf_aligned_total: Total number of SPF algined messages.
• dmarc_spf_pass_total: Total number of messages with raw SPF pass.
• dmarc_dkim_aligned_total: Total number of DKIM algined messages.
• dmarc_dkim_pass_total: Total number of messages with raw DKIM pass.

Each of these metrics is subdivided by the following labels:

• reporter: Domain from which a DMARC aggregate report originated.
• from_domain: Domain from which the evaluated email originated.
• dkim_domain: Domain the DKIM signature is for.
• spf_domain: Domain used for the SPF check.

In addition, there is a dmarc_invalid_reports_total metric with a count of DMARC report emails from which no report could be parsed. It is subdivided by a single from_email label.

Installation

This describes the manual setup fo dmarc-metrics-exporter. An Ansible role for automated deployment is provided in roles. Further instructions for Ansible are given in the readme file provided in that directory.

It is best to run dmarc-metrics-exporter under a separate system user account. Create one for example with

adduser --system --group dmarc-metrics

Then you can install dmarc-metrics-exporter with pip from PyPI for that user:

sudo -u dmarc-metrics pip3 install dmarc-metrics-exporter

You will need a location to store the metrics.db that is writable by that user, for example:

mkdir /var/lib/dmarc-metrics-exporter
chown dmarc-metrics:dmarc-metrics /var/lib/dmarc-metrics-exporter

Configuration

To run dmarc-metrics-exporter a configuration file in JSON format is required. The default location is /etc/dmarc-metrics-exporter.json.

Because the configuration file will contain the IMAP password, make sure to ensure proper permissions on it, for example:

chown root:dmarc-metrics /etc/dmarc-metrics-exporter.json
chmod 640 /etc/dmarc-metrics-exporter.json

An example configuration file is provided in this repository in config/dmarc-metrics-exporter.sample.json.

The following configuration options are available:

• listen_addr (string, default "127.0.0.1"): Listen address for the HTTP endpoint. Use "0.0.0.0" if running in a dockerized environment.
• port (number, default 9797): Port to listen on for the HTTP endpoint.
• imap (object, required): IMAP configuration to check for aggregate reports.
  • host (string, default "localhost"): Hostname of IMAP server to connect to.
  • port (number, default 993): Port of the IMAP server to connect to.
  • username (string, required): Login username for the IMAP connection.
  • password: (string, required): Login password for the IMAP connection.
  • use_ssl: (boolean, default true): Whether to use SSL encryption for the connection. Disabling this will transmit the password in clear text! Currently, there is no support for STARTTLS.
  • verify_certificate: (boolean, default true): Whether to verify the server's SSL certificate. You might have to set this to false if you are using a self-signed certificate. If this is disabled, someone else could impersonate the server and obtain the login data.
• folders (object):
  • inbox (string, default "INBOX"): IMAP mailbox that is checked for incoming DMARC aggregate reports.
  • done (string, default "Archive"): IMAP mailbox that successfully processed reports are moved to.
  • error: (string, default "Invalid"): IMAP mailbox that emails are moved to that could not be processed.
• storage_path (string, default "/var/lib/dmarc-metrics-exporter"): Directory to persist data in that has to persisted between restarts.
• poll_interval_seconds (number, default 60): How often to poll the IMAP server in seconds.
• deduplication_max_seconds (number, default 604800 which is 7 days): How long individual report IDs will be remembered to avoid counting double delivered reports twice.
• logging (object, default {}): Logging configuration, see the "Logging configuration" section below.

Logging configuration

When providing a custom logging configuration, it must follow the dictionary schema (version 1) described in the logging.config documentation. In general, a provided top-level key will replace the default configuration, but there are some exceptions. The following keys are always fixed:

• version will always be 1.
• incremental will always be false.
• formatters is fixed and provides the following formatters:
  • plain renders human-readable log messages without colors.
  • colored renders human-readable log messages with colors.
  • json renders structured JSON log messages.

In addition, the root key has some special handling. If it is overridden, but not handlers key is provided, handlers: ['default'] will be inserted automatically. Also, the level key will be set to 'DEBUG' if the application is started with the --debug flag.

Configuring log level

To change the log level globally:

{
    "logging": {
        "root": {
            "level": "WARNING"
        }
    }
}

Configuring logging format

To change the logging format:

{
    "logging": {
        "handlers": {
            "default": {
                "class": "logging.StreamHandler",
                "formatter": "json"
            }
        }
    }
}

Valid formats are plain, colored, and json.

Disabling Uvicorn access logs

To disable the Uvicorn access logs:

{
    "logging": {
        "loggers": {
            "uvicorn.access": {
                "propagate": false
            }
        }
    }
}

Usage

To run dmarc-metrics-exporter with the default configuration in /etc/dmarc-metrics-exporter.json:

sudo -u dmarc-metrics python3 -m dmarc_metrics_exporter

To use a different configuration file:

sudo -u dmarc-metrics python3 -m dmarc_metrics_exporter --configuration <path>

You can enable debug logging with the --debug if you do not want to provide your own logging configuration:

sudo -u dmarc-metrics python3 -m dmarc_metrics_exporter --debug

systemd

Instead of manually starting the dmarc-metrics-exporter, you likely want to have it run as a system service. An example systemd service file is provided in this repository in config/dmarc-metrics-exporter.service. Make sure that the paths and user/group names match your configuration and copy it to /etc/systemd/system to use it. To have systemd pick it up a systemctl daemon-reload might be necessary.

You can than start/stop dmarc-metrics-exorter with:

systemctl start dmarc-metrics-exporter
systemctl stop dmarc-metrics-exporter

To have dmarc-metrics-exporter start on system boot:

systemctl enable dmarc-metrics-exporter

Docker

A new docker image is build for each release with GitHub Actions as described in this yaml-file: .github/workflows/docker-publish.yml.

Note that you should configure the listen_addr to 0.0.0.0 to be able to access the metrics exporter from outside the container.

Example docker-compose file:

version: "3"

services:

  dmarc-metrics-exporter:
    # source: https://github.com/jamborjan/dmarc-metrics-exporter/pkgs/container/dmarc-metrics-exporter
    container_name: dmarc-metrics-exporter
    hostname: dmarc-metrics-exporter
    image: jgosmann/dmarc-metrics-exporter:1.2.0
    restart: unless-stopped
    user: 1000:1000 #PUID=1000:PGID=1000
    expose:
      - 9797
    volumes:
      - '/host/folder/dmarc-metrics-exporter.json:/etc/dmarc-metrics-exporter.json'
      - '/host/folder/dmarc-metrics-exporter/metrics:/var/lib/dmarc-metrics-exporter:rw'
    logging:
      driver: "json-file"
      options:
        tag: "{{.ImageName}}|{{.Name}}|{{.ImageFullID}}|{{.FullID}}"
    networks:
      - YourDockerLan

# $ docker network create -d bridge --attachable YourDockerLan
networks:
  YourDockerLan:
    external:
      name: YourDockerLan

Prometheus

Example prometheus config file:

global:
  scrape_interval: 15s
  evaluation_interval: 15s

rule_files:

scrape_configs:

  - job_name: 'dmarc-metrics-exporter'
    static_configs:
      - targets: ['dmarc-metrics-exporter:9797']

Grafana

An example configuration file is provided in this repository in config/dmarc-metrics-exporter.grafana.sample.json. This example dashboard displays the collected metrics as shown in the screenshot below.

Example grafana dashboard

Hints

You should not use your normal email and password credentials for the dmarc-metrics-exporter. If you are not able to create a dedicated service account email account, you should use an app password.

Microsoft Exchange Online

• App passwords are available when you are using Multi Factor Authentication (MFA). Manage app passwords for two-step verification
• If you don't see the app passwords option or get an error, check if MFA is enabled for the user.
• If you still don't see the app passwords option, check if app passwords are allowed in your organization
• Finally, ensure that IMAP is enabled for the user.

Development

Prerequisites

• Python
• pre-commit
• Poetry
• Docker

Setup development environment

pre-commit install
poetry install

Run tests

docker-compose up -d
poetry run pytest