prometheus/alertmanager

Alertmanager is a companion tool to Prometheus, which is a widely-used system for monitoring servers, applications, and infrastructure. When Prometheus detects that something is wrong, such as a server running low on memory or a service responding too slowly, it sends alerts to Alertmanager. Alertmanager then decides what to do with those alerts: where to send them, how to group related ones together, and when to suppress them to avoid flooding on-call teams with redundant noise. The three main things Alertmanager handles are routing, grouping, and silencing. Routing means directing alerts to the right destination based on their labels, such as sending database alerts to the database team and critical alerts to a pager service. Grouping means batching multiple alerts that share the same cause into a single notification, so one database outage does not generate dozens of separate pages. Silencing allows you to mute alerts during planned maintenance windows or while a known issue is already being worked on. Supported notification destinations include email, PagerDuty, OpsGenie, and a generic webhook receiver that can connect to almost anything else. Configuration is done through a YAML file that defines routes as a tree of rules with labels as matching criteria. The README includes a detailed example configuration covering multiple teams, severity levels, and grouping strategies. Alertmanager ships as a precompiled binary for most platforms and is also available as a Docker image. A command-line tool called amtool is bundled with every release and lets you view active alerts, create silences, check routing decisions for hypothetical alerts, and manage the cluster state if you are running multiple Alertmanager instances for redundancy. The API is version 2, follows the OpenAPI specification, and is accessible under the /api/v2 path. The project is part of the official Prometheus organization and follows the same open-source governance as the rest of the ecosystem.