Monitoring stack

Uptime checks, TLS expiry, server metrics, dashboards and Telegram alerts in one docker compose up. Know your site is down before your customers do.

Maintained by GonService · Português

flowchart LR
    BB[Blackbox exporter<br/>HTTP + TLS checks] --> P[(Prometheus<br/>metrics + alert rules)]
    NE[Node exporter<br/>CPU, memory, disk] --> P
    P --> G[Grafana<br/>Overview dashboard]
    P --> AM[Alertmanager<br/>grouping, inhibition]
    AM --> T[Telegram]
    BB -.probes.-> S[Your sites and APIs]

Loading

What you get

Component	Version	Role
Prometheus	2.54.1	Scrapes metrics every 30 s, evaluates alert rules, keeps 30 days
Blackbox exporter	0.25.0	Checks each URL in targets/http.yml: status, latency, TLS expiry
Node exporter	1.8.2	CPU, memory, disk and filesystem metrics of the host
Alertmanager	0.27.0	Groups alerts, suppresses noise, sends to Telegram (resolved too)
Grafana	11.2.2	Provisioned data source and Overview dashboard: status, response time, uptime, TLS days left, CPU, memory, disk

Alerts

Alert	Fires when	Severity
SiteDown	HTTP check fails for 2 minutes	critical
SiteSlow	Average response time above 2 s for 5 minutes	warning
TLSCertificateExpiringSoon	Certificate expires in less than 14 days	warning
TargetDown	A monitored component stops responding	critical
HighCPU / HighMemory	Above 90% for 10 minutes	warning
DiskAlmostFull	Less than 10% free	critical
DiskWillFillIn24h	Current trend fills the disk within a day	warning

Rules have unit tests (prometheus/tests) run by make check and in CI. A site that is down does not also page you for being slow (inhibition rule).

Run it

make init    # creates .env and secrets/telegram_bot_token
# edit .env: GRAFANA_ADMIN_PASSWORD and TELEGRAM_CHAT_ID
# put the bot token (from @BotFather) in secrets/telegram_bot_token
make up      # Grafana on http://localhost:3000

• Monitor your URLs: edit prometheus/targets/http.yml; Prometheus picks up the change within a minute.
• More servers: run node-exporter on each one and add it to the node job in prometheus.yml, then make reload.
• Without Telegram: until TELEGRAM_CHAT_ID and the bot token are set, Alertmanager starts with alertmanager.yml and alerts only show in Alertmanager and Grafana.
• Slack or e-mail instead of Telegram: replace the receiver in telegram.yml with slack_configs (api_url_file) or email_configs (auth_password_file), keeping secrets in the secrets/ folder.
• Ports are bound to 127.0.0.1. To expose Grafana, put it behind a reverse proxy with HTTPS.

Em português

Monitoramento completo com um docker compose up: verificação de disponibilidade e tempo de resposta dos seus sites e APIs, aviso de certificado TLS vencendo, métricas de CPU, memória e disco, painel no Grafana e alertas no Telegram. As regras de alerta têm testes automatizados. Para usar: make init, preencha o .env e o token do bot, e make up.

Quer monitoramento e alertas no seu sistema? A GonService instala e mantém: gonserviceit.com.br/devops · contato@gonserviceit.com.br

License

MIT