If a device can send a reading over the internet, NoDisrupt can monitor it: chart what it measures, alert you when a value leaves the range you set, and raise an incident when the device stops reporting at all.
That is the whole idea, and it is deliberately broad. The same mechanism that watches a web server watches a temperature probe in a boiler room, a Raspberry Pi in the field, or a power supply in a cabinet. What changes is the reading, not the tool.
A connected device fails in two ways
It reports a bad value, or it stops reporting. These are not the same problem, and most setups only catch the first.
A reading that drifts out of range is visible on a dashboard, if someone happens to be looking at the moment it does. A device that goes silent shows up as nothing: no alert, no red, no change, just an absence that is easy to mistake for calm. A dead battery, a tripped breaker and a dropped connection all look identical to a chart that simply stops updating. The second failure is the dangerous one precisely because nothing turns red on its own.
Watching a device properly means catching both, and treating the silence as seriously as the bad number.
What counts as a device
Anything that measures something and can reach the internet. A few of the common ones, and the page that covers each in detail.
| Device or sensor | What you would watch |
|---|---|
| Temperature and humidity sensors | A cold room, a rack or a store drifting out of its safe range. |
| A server room or rack | The ambient temperature where the hardware actually lives. |
| Fridges and freezers | Stock spoiling quietly overnight because a door was left ajar. |
| A Raspberry Pi or single-board computer | A field unit running hot, or one that has simply stopped. |
| UPS and power supplies | Mains lost, or a battery quietly running down to nothing. |
| A server or host | CPU, memory and disk on the machine itself. |
| Any device with a controller or gateway | Tank level, pump state, a door open, signal strength. |
The list is not meant to be exhaustive. The test is simpler than any category: if the thing produces a reading and can send it over a network, it can be watched here.
How it works, in three moves
The shape is the same whatever the device is.
For the exact mechanics of getting a reading in, the IoT device monitoring page and the custom metrics check both work through a concrete example end to end.
Alerting on silence is the point
The single setting that earns its keep is how long a device may be quiet before it counts as offline. That one choice catches the failures a dashboard never will: the dead battery, the power cut, the gateway that dropped off the network at three in the morning.
Catching it is only half the job. The alert then has to reach a person who can act, and if the first person does not answer, it has to go somewhere else rather than sit unread. That routing is the work of escalation policies. For the specific case of mains power dropping out, the power outage alerts page covers the setup directly.
Show it to the people who depend on it
The people who care whether a device is healthy are often not the ones watching the dashboard. A status page puts any monitored device in front of them, so a colleague or a customer can see at a glance whether it is reporting, without having to ask.
Common questions
Can I monitor a temperature sensor with NoDisrupt?
Yes. A sensor that can report its reading over the internet can feed NoDisrupt, which charts the temperature, alerts you when it leaves the range you set, and raises an incident if the sensor stops reporting at all. The same holds for humidity, power, and other environmental readings.
What kinds of device can report in?
Anything that can send a reading over an internet connection: environmental sensors, single-board computers such as a Raspberry Pi, power supplies, and industrial equipment with a controller or gateway in front of it. The device does the measuring; NoDisrupt does the watching and the alerting.
Do I need to install an agent or SDK on the device?
No. For a connected device that already reports a reading, nothing is installed on the device beyond what it takes to send that reading over the internet. An agent is only involved when you want host-level figures such as CPU, memory and disk from a server you run.
What happens when a device goes offline?
That is treated as an incident in its own right. You choose how long a device may be silent before it counts as offline, and once it crosses that window NoDisrupt alerts a human the same way it would for a bad reading. A dead sensor is the failure dashboards usually miss.
Can I monitor a whole fleet of devices?
Yes. Each device is watched on its own, so a leaked key or a single failure on one does not affect the others. A fleet counts against your plan's monitor limit, so the practical question is how many devices you need to watch at once, not whether it is possible.
Can customers or colleagues see device status?
Yes. Any device being monitored can be added to a public status page, so the people who depend on it can see whether it is reporting without asking you.
Starting from the software side instead? The uptime monitoring guide covers watching services and endpoints, and IoT device monitoring has the device-side detail.