Monitoring & alerts

What the agent actually reports, and how to get notified when something's wrong.

What's collected

Pushed every 15 seconds by the agent, no configuration needed:

  • CPU, memory, disk, network, load average — full time-series history (retention varies by plan, see Plans & billing)
  • Uptime / online-offline status
  • Top processes snapshot
  • Per-partition disk usage
  • Disk health / SMART status (Linux via smartmontools, Windows via Get-PhysicalDisk)
  • Docker containers, if Docker is detected on the box
  • Network interfaces and listening ports

All of this is viewable live on the server's detail page, and pullable via the read-only API.

Alert rules

An alert rule watches one metric on one server and fires when a condition holds for a minimum duration:

MetricNotes
offlineNo metrics push received recently
cpu / memory / diskThreshold + condition (greater than / less than) + duration, e.g. "CPU > 90% for 5 min"
processA named process disappearing (or appearing) from the top-processes list
backup_failedThe most recent backup run for this server ended in failure
monitor_downAny external monitor on this server (see below) reports down

A short duration requirement debounces single-blip noise — a CPU rule set to "5 minutes" won't fire on a one-second spike.

Delivery channels

ChannelSetup
EmailAlways available — no setup. Extra recipient emails can be added per rule.
TelegramLink your Telegram account from Settings; alerts arrive as bot messages.
SlackPaste an incoming-webhook URL from Settings — one webhook covers your whole account.
Generic webhookSet per alert rule — MonitorFree POSTs a JSON payload to any URL you provide when the rule fires.

Incident timeline

Each time an alert condition transitions from OK → firing (and back), that's logged as an alert event with a timeline you can review later — useful for "when exactly did this start" during a postmortem. Team members can leave comments on an event for shared incident notes.

Every server gets four default alerts on creation (offline, CPU>90%/5min, backup failed, monitor down) so you're never accidentally running with zero alerting configured.