alertmanager

Start Alertmanager with default configuration

$ alertmanager --config.file=[alertmanager.yml]

$ alertmanager --config.file=[alertmanager.yml] --storage.path=[/var/lib/alertmanager]

$ alertmanager --config.file=[alertmanager.yml] --cluster.peer=[peer1:9094] --cluster.peer=[peer2:9094]

$ alertmanager --config.file=[alertmanager.yml] --config.check

$ alertmanager --config.file=[alertmanager.yml] --web.listen-address=[0.0.0.0:9093]

alertmanager [options]

Alertmanager handles alerts sent by Prometheus and other monitoring systems. It manages deduplication, grouping, silencing, inhibition, and routing of alerts to notification receivers such as email, Slack, PagerDuty, and webhooks.
When Prometheus detects a condition matching an alerting rule, it sends alerts to Alertmanager. Alertmanager groups related alerts together, waits for configured intervals to batch notifications, and routes them to appropriate receivers based on label matching rules.
Key features include silences to mute alerts during maintenance, inhibition rules to suppress alerts when related alerts are firing, and high availability through a gossip-based cluster protocol that ensures alerts are not duplicated across instances.
The web UI at the configured listen address provides alert management, silence creation, and cluster status visibility.

--config.file path

Path to the Alertmanager configuration file.

Directory for storing notification state and silences.

Address to listen on for the web UI and API (default: :9093).

External URL for generating links back to Alertmanager.

Address for cluster communication (default: 0.0.0.0:9094).

Initial peer addresses for cluster formation (repeatable).

Address to advertise in the cluster.

Log level: debug, info, warn, error.

Log format: logfmt or json.

Validate configuration and exit.

Alertmanager requires a properly configured YAML file; invalid configuration prevents startup. Cluster mode requires all peers to be mutually reachable on the cluster port. Notification state is stored locally; losing storage can cause duplicate notifications. Time-based grouping may delay critical alerts if intervals are set too long.

Alertmanager was developed by SoundCloud as part of the Prometheus ecosystem, first released in 2013. It became a graduated project of the Cloud Native Computing Foundation (CNCF) alongside Prometheus in 2018. The tool has evolved to support numerous notification integrations and high-availability deployments in modern cloud-native infrastructure.

prometheus(1), amtool(1), grafana(1)