Power outage alert for a server room: how to know the mains failed before the UPS runs out
The most reliable power outage alert for a small server room is a dry-contact output on the UPS, wired to a network logger that itself runs on UPS power and sends e-mail the moment the contact changes. It works when the UPS network card is absent, when the monitoring server is on the failed circuit, and when the only thing still alive in the room is the UPS and the switch it feeds.
In short
- Power is the cause of 45 % of serious outages (Uptime Institute 2025), most involving the UPS. Knowing within a minute that you are on battery is the difference between a clean shutdown and a corrupted array.
- Most UPS units have a relay or dry-contact output for "on battery" and "low battery". Wire it to a logger input; no UPS software, no SNMP card needed.
- Put the PoE switch and the logger on the UPS. The logger then reports through the outage and adds the alarm that follows every power cut: rising temperature when the cooling has stopped.
- Alert on two channels: e-mail from the device and push or SMS via a cloud app, because the mail server may be on the same failed circuit.
Why the UPS software alone is not enough
Every UPS ships with software: a USB agent on a server, or a network card with SNMP and e-mail. Both have the same weakness. The agent runs on a server that may be the first thing shut down, and its e-mail goes through a mail relay that may be on the same failed circuit. The network card is an optional extra on small units, costs a significant fraction of the UPS, and, as the Home Assistant and Eaton NUT threads on the forums show, integrations with it break after software updates with some regularity.
The failure is rarely dramatic. The mains drops at 21:15, the UPS carries the load, the agent sends a mail that queues on a relay without power, the battery empties at 21:40, the NAS loses power mid-write, and the first human learns of it at 07:30 from a "volume degraded" message. The fix is an alert path with as few dependencies as possible.
The dry contact: the alert that survives everything
Most UPS units from 1 kVA upwards, and many smaller ones, have a relay output on the back: a terminal block or a DB9 "dry contact" port with pins that close or open on "on battery", "low battery" and "UPS fault". It is the oldest interface there is, needs no driver, and works while the UPS has any charge left. Check the manual: Eaton, APC and Riello all document the pinout, and on many models the port is shared with the RS-232 signalling.
A dry contact is exactly what a logger's alarm input expects. The logger polls the input, and when the contact changes state it records the event with a timestamp and sends its e-mail. There is nothing to configure on the UPS side beyond, on some models, a jumper or menu setting choosing which condition drives the relay.
| Signal | Meaning | Action |
|---|---|---|
| On battery | Mains failed, UPS carrying load | Immediate alert; start the clock |
| Low battery | Minutes of runtime left | Trigger shutdown, escalate |
| UPS fault / bypass | UPS not protecting the load | Ticket, no waiting for an outage to find out |
Wiring it to a logger
- Power the logger from the UPS. A PoE logger such as EnviSensor ETH PoE draws power from the switch; put the switch on the UPS and the logger is covered. A mains-powered logger goes on a UPS outlet directly.
- Wire "on battery" to input 1. Two cores from the UPS relay terminals to the logger's dry-contact input. Note in the configuration whether the contact is normally open or normally closed; the manual says which.
- Wire "low battery" to input 2 if the UPS offers it and you have inputs to spare. Otherwise, "on battery" plus a timer in your head is workable.
- Set alarms. E-mail from the device on input change, to a group that includes at least one phone. If the logger reports to a cloud app, enable push there as a second channel.
- Test it. Pull the UPS mains plug for thirty seconds. Confirm the e-mail arrives, the push arrives, and the monitoring platform raised an event over SNMP. Repeat once a quarter.
For a controller with more inputs and relay outputs, such as an EnviSwitch, the same contact can also drive an output: a beacon in the corridor, or a relay that starts a scripted shutdown.
The second alarm: temperature after a power cut
In a small server room the air conditioning is almost never on the UPS. When the mains fail, cooling stops while the servers keep running on battery, and the room can climb 1 °C per minute in a closed closet with a couple of kilowatts of load. The UPS may bridge the outage; the heat that builds up during it is what actually reboots the switch twenty minutes after the power is back.
This is why the power alert belongs on the same logger as the temperature probe. One e-mail says "on battery since 21:15". The next says "top of rack 31 °C" at 21:27. Together they tell the on-call person whether they have time to wait for the mains or need to shut down now. The Uptime Institute lists cooling as the second cause of serious outages at 19 %; a good share of those follow a power event that nobody was watching.
If the UPS has no contact output
Small desktop UPS units often have USB only. Two workarounds. The first is a mains-monitoring relay: a DIN-rail relay with a 230 V coil, plugged into a non-UPS socket, whose contact drops when the mains fail. Wire that contact to the logger input. It costs a few euros and turns any socket into an "is the mains on" sensor. The second is a plug-in power adapter for a non-UPS socket feeding a logger input via a small relay; the same idea in a plastic box.
In both cases the logger itself must be on UPS power, or it disappears with the mains and the alert is a silent "device unreachable" in the monitoring, which is still better than nothing but only if someone has configured that alert too.
Frequently asked questions
How can I get an alert when the power fails in my server room?
Wire the UPS's dry-contact "on battery" output to an input on a network logger that is itself powered from the UPS, and have the logger send e-mail and push notifications on the contact change. It works without UPS software or a network card.
What is a UPS dry contact?
A relay output on the UPS, usually a terminal block or DB9 port, that closes or opens on conditions such as on battery, low battery or fault. It carries no voltage of its own and connects directly to an alarm input.
Why not just use the UPS software or SNMP card?
They work until the server running the agent shuts down or the mail relay is on the failed circuit, and network cards are an extra cost that integrations break on. A dry contact to an independently powered logger has no such dependencies.
My UPS has only a USB port. What then?
Use a mains-monitoring relay in a non-UPS socket: its contact drops when mains power fails, and that contact goes to the logger input. The logger itself must be on UPS power.
Why monitor temperature together with power?
Because cooling is rarely on the UPS. After a power cut the room heats up while servers run on battery, and the temperature alarm tells you whether to wait or shut down. One logger can do both.
Related pages
EnviSensor ETH PoE
Runs on the UPS through the PoE switch; dry-contact inputs.
EnviSwitch
8 inputs and 8 relay outputs for signals and control.
Rack temperature limits
Why the UPS battery is the first thing heat kills.
Need monitoring at your site?
We will help you choose sensors and the way they run — locally on your LAN, in your own monitoring system or in the CE MonitorApp cloud.
