MONITORING

Every server's vitals, one glance.

Real-time CPU / memory / disk / network / swap / load streaming straight from the agent over mTLS. Configurable refresh from 1s to 5s, 30-day metric history, process-level attribution.

cloudnan.com / dashboard / monitoring

Fleet Monitoring

3 servers · refresh 5s · 30-day history

web-01.cloudnan.dev

cpu

42%

mem

67%

Disk

38%

db-01.cloudnan.dev

cpu

78%

mem

84%

Disk

71%

cache-01.cloudnan.dev

cpu

12%

mem

23%

Disk

db-01 memory above 80% for 8 minutes

Routed to: Telegram, Email · cooldown 15 min

How it works

What happens under the hood.

01
Agent reads /proc + sysfs
Lightweight collection — no external monitoring stack to install or run.
02
Streams via mTLS gRPC
Outbound-only connection from agent to control plane; no inbound ports needed.
03
Panel renders live charts
Sparklines for CPU / mem / disk / net update on every server card; click any to drill into the full metric.
04
Process list with kill
See top CPU / memory consumers per agent. Kill a runaway process from the panel — audit-logged.
05
Alerts fire on threshold
CPU >90% for 5 min, disk >85%, agent offline 3 min — every threshold notifies via your wired channels.
06
30-day history
Time-series storage on the control plane. Zoom out to see weekly cycles; zoom in for incident review.

What you get

The capabilities that ship.

CPU / memory / disk / network

Per-core CPU breakdown, swap usage, per-mount disk, per-interface network — every dimension you'd open htop for.

Process list

Top consumers ranked by CPU or memory. Kill, renice, or send signal — gated through the action audit log.

Custom alert rules

Per-agent thresholds with cooldown windows. Route to Email / Telegram / Slack / Discord / WhatsApp / webhook.

Configurable refresh

Default 5s update; tighten to 1s for incident response. Pause when the page is in the background.

30-day retention

Time-series data persisted on the control plane. Replay any past 30 days at full resolution.

Multi-server overview

Fleet dashboard aggregates the whole roster. Sort by hot-load, alert state, or last-seen-online.

FAQ

Common questions.

How is this different from Prometheus / Grafana?

Same metric surface, much less ops overhead. No prometheus.yml to write, no Grafana dashboard to author, no exporters to run. For deeper analytics teams, Cloudnan can also export metrics to your existing stack via a webhook.

Does it work with Docker stats?

Yes — per-container CPU / memory / network are surfaced on the Docker page. Same data path: agent reads cgroups, streams over mTLS.

What about custom metrics from my app?

Application-level metrics aren't in the default scope yet. Today you'd push them to Prometheus or Datadog and watch there. We're evaluating an OTLP endpoint for late 2026.

Related capabilities

Notifications & Alerts

Wire alert rules to email / Telegram / Slack / Discord.

Read

Drift Detection

Know when a config that affects monitoring changes.

Read

AI Operations Assistant

Ask the AI to investigate a metric spike.

Read