5. Dashboards & Alerts

[ksobrenat32]

Import kube-prometheus default dashboards in Grafana.
Build custom dashboards:

• Mirrors: probe latency (from blackbox), requests per sec (from nginx exporter), bytes per IP (from ES), location.
• Tor: bandwidth metrics from MetricsPort.

Example alert rule:

groups:
- name: mirror-alerts
  rules:
  - alert: MirrorDown
    expr: probe_success{job="mirror-servers"} == 0
    for: 5m
    labels:
      severity: critical
    annotations:
      description: "Mirror {{ $labels.instance }} is unreachable"

Research:

• Writing PromQL queries for error rate, traffic spikes, availability.
• Alertmanager routes (email, Slack, Telegram, etc).
• Grafana alerting vs Prometheus alerting (when to use each).