alertmanager-dashboard

This project provides a simple, self-hosted solution for monitoring active alerts from Prometheus Alertmanager. It consists of two independent Go binaries, each packaged as a separate container image, allowing for flexible deployment. **Components:** 1. **Receiver:** This binary acts as a webhook endpoint for Alertmanager. It listens for POST requests at `/webhook`, processes incoming alert data, and stores the current state in a Valkey (Redis-compatible) database. It specifically handles `Watchdog` alerts by updating a last-seen timestamp without treating them as active issues. 2. **Dashboard:** This binary reads the state from Valkey and serves a web-based HTML dashboard. It also exposes a `GET /api/state` endpoint for programmatic access to the current alert status. The dashboard can be configured to highlight stale updates if no new webhooks have been received within a specified threshold. **State Management:** Active alerts are stored in a Valkey hash keyed by their unique fingerprint. Resolved alerts are automatically removed. If a fingerprint is missing, it is derived from sorted labels to ensure consistency. **Configuration:** Both binaries are configured via environment variables, including: - `LISTEN_ADDR`: HTTP listen address (default `:8080`). - `VALKEY_ADDR`: Address of the Valkey/Redis instance. - `VALKEY_USERNAME` / `VALKEY_PASSWORD`: Authentication credentials. - `STALE_AFTER`: Threshold for marking updates as stale (default `90m`). **Deployment:** The project supports Docker Compose for local development and testing. Published container images are available on GitHub Container Registry (`ghcr.io`). The final images are optimized for security and size, using `scratch` as the base, running as non-root (UID/GID 1000), and including binary-backed health checks.