What is the NIZU Uptime Monitor?

The Uptime Monitor (plugin key NizuUptime) watches whether the websites, servers and services a company depends on are actually up, and tells someone by email or Telegram the moment one of them isn't.

In one sentence: Uptime Monitor answers "is it up right now, and how reliable has it been" for every host you register, without needing an agent installed on most of them.

How it checks a host

Each monitored server has a type, and the type decides how it is checked:

• HTTP / HTTPS a cURL request against the host (with an optional port); 2xx–3xx counts as up, anything else a timeout, a connection error, a 4xx/5xx counts as down.
• TCP port opens a socket to the host and port; a successful connect is up, a refused or timed-out one is down.
• Ping (ICMP) a single system ping with a short timeout.
• Reverse heartbeat for a machine with no public IP the check direction is flipped: a small script installed on the machine calls the workspace, not the other way round. See below.

Every check is timed, and its outcome up, down, the response time in milliseconds, and any error text is written to a log row. The server's own status and last_checked fields always reflect the most recent check only; the history in the log is what makes an uptime percentage possible.

The reverse heartbeat, the plugin's signature feature

Most uptime tools assume they can reach the thing they monitor. A laptop behind NAT, a machine on a residential connection, or a server with firewalled inbound traffic breaks that assumption. Reverse heartbeat solves it by turning the check around: the Uptime Monitor generates a unique token per server (an MD5 of its ID) and a ready-to-paste bash script plus a systemd unit. Installed on the target machine, the script checks its own service locally every 60 seconds and POSTs the result to /nizuuptime/heartbeat/<token> on the workspace. As long as reports keep arriving, the server reads as up. If the scheduled check interval elapses with nothing received, the server is marked down the absence of a heartbeat is the failure signal.

Alerts

When a server's status changes up to down, or down to up an alert can go out by email, Telegram, or both, per server, on top of the global notification switches in Settings. Nothing is sent for a status that hasn't changed, so a server that stays down doesn't spam the same channel on every check.

Who it is for

• Administrators and IT staff who run the company's own infrastructure and want a first line of alerting without standing up a separate monitoring stack.
• Managers who want the "is everything OK" answer on the dashboard without opening a monitor at all.
• AI agents connected to the workspace, which can read uptime and response-time statistics through a dedicated, read-only reporting tool never change a monitor's configuration.

Key capabilities

• HTTP, HTTPS, TCP and Ping checks, plus reverse heartbeat for unreachable machines.
• A per-server check interval from 1 minute to 1 hour.
• Email and Telegram alerts, configured globally and toggled per server.
• A server list with live status, a 10-segment uptime bar, response time, and a one-click enable/disable toggle.
• A full check log per server, exportable, with automatic retention-based purging.
• Dashboard widgets: a servers-status card plus four KPI tiles (servers online, servers down, 24-hour uptime, average response time).
• Role-based permissions, separate for managing servers and for reading logs.
• A read-only reporting tool for AI agents, so uptime can be asked about in plain language.

Frequently asked questions

Does it need an agent installed on every server?

No HTTP, HTTPS, TCP and Ping checks are all made remotely from the workspace. An agent (the reverse-heartbeat script) is only needed for a machine the workspace cannot reach directly.

Can an AI agent turn a monitor off?

No. The only tool exposed to agents is a reporting tool; it cannot create, edit, delete or pause a server.