Skip to main content

Monitoring

Prometheus and Grafana run both locally (docker-compose) and in the cluster (infra/k8s/prometheus.yaml, infra/k8s/grafana.yaml), with the same dashboards, alerting rules, and Telegram integration. In K8s, config is mounted from ConfigMaps instead of bind-mounted files.

Local (docker-compose)

  • Grafana: http://localhost:8080/grafana/ — default login admin / admin. Dashboards and alerting rules are auto-provisioned.
  • Prometheus: http://localhost:8080/prometheus/ — gated by HTTP basic auth. Default login is admin / devpulse; override via PROMETHEUS_AUTH_USER/PROMETHEUS_AUTH_PASSWORD in infra/.env. The credential file itself is generated at container startup, never committed.

Kubernetes

The Azure VM sizing dashboard estimates RAM from our own app-level metrics (JVM memory, py-intelligence resident memory), so it works in both local and in-cluster Grafana.