Icinga Notifications to Prometheus Alertmanager

Send Icinga2 notifications to the Prometheus Alertmanager.

It transforms the Icinga2 notification into JSON data and sends this data to the Alertmanager API: https://prometheus.io/docs/alerting/latest/alerts_api/

Other Icinga and Prometheus integrations we provide:

• https://github.com/NETWAYS/check_prometheus/
• https://github.com/NETWAYS/icinga2-exporter
• https://github.com/NETWAYS/icingaweb2-module-perfdatagraphs-prometheus
• https://github.com/NETWAYS/alertmanager-icinga-bridge

Installation

Requirements:

• curl

Usage

Usage: notify-alertmanager.sh [OPTIONS]

Options:
  -t TYPE              Object type: host | service
  -T NOTIFICATION_TYPE Icinga notification type: PROBLEM | RECOVERY | ACKNOWLEDGEMENT | FLAPPINGSTART | FLAPPINGSTOP | DOWNTIMESTART | DOWNTIMEEND
  -H HOST_NAME         Icinga host name ($host.name$)
  -u ALERTMANAGER_URL  Alertmanager base URL (default: http://localhost:9093)
  -c COMMENT           Notification comment
  -a AUTHOR            Notification author
  -i ICINGA_URL        Icinga URL for the object (used in annotations)
  -l LABELS            Extra labels as comma-separated key=value pairs
  -s STATE             Object state: UP | DOWN | UNREACHABLE | OK | WARNING | CRITICAL | UNKNOWN
  -v                   Verbose output
  -h                   Show this help

Host options:
  -d DISPLAY_NAME Host display name ($host.display_name$)
  -A ADDRESS      Host address ($host.address$)

Service options:
  -n SERVICE_NAME         Service name ($service.name$)
  -N SERVICE_DISPLAY_NAME Service display name ($service.display_name$)

Examples:

notify-alertmanager.sh -t service -T PROBLEM -H myhost \
-d "My Host" -A 192.168.1.1 -s CRITICAL -n "ping" -N "Ping Check" -u http://localhost:9093 -l severity=critical

notify-alertmanager.sh -t service -T RECOVERY -H myhost \
-d "My Host" -A 192.168.1.1 -s CRITICAL -n "ping" -N "Ping Check" -u http://localhost:9093 -l severity=critical

Note that the Icinga notification type (PROBLEM, RECOVERY, etc.) and state (OK, CRITICAL, etc.) are not used a labels in the final alert. This is because the Prometheus Alertmanager uses labels to match firing-resolved alerts, using changing type/state as labels would hinder this matching.

Use the -l flag to add additional static labels to add further information: -l severity=critical.