UPS Software: What It Does and Which Tool to Use

What UPS software actually does

A UPS knows a lot about the power situation: whether it is on utility or battery, how charged the battery is, how many minutes the current load can run, and whether the battery has failed a test. The computer it powers knows none of that unless something reads it. UPS software is that something. Its jobs, roughly in order of importance:

1. Graceful shutdown. When the battery reaches a threshold (a charge percentage, a runtime estimate, or a fixed time on battery), the software tells the operating system to close applications, flush disks and halt. This is the feature that prevents corrupted databases, damaged file systems and half-written virtual machine disks.
2. UPS power-off and restart. After the OS halts, the software (or the shutdown scripts it installs) can tell the UPS to turn its outlets off after a delay. When utility power comes back, the UPS restores output, and machines set to power on after AC loss boot themselves.
3. Notifications. Email, desktop pop-ups, syslog entries, SNMP traps or webhook scripts when power fails, returns, or the battery needs replacing.
4. Logging. A history of transfers to battery, input voltage events and self-test results. This is how you discover that the UPS has been switching to battery 30 times a day because of voltage sags.
5. Control and diagnostics. Starting self-tests and runtime calibration, adjusting transfer voltages and sensitivity, muting the alarm, and reading the battery install date.

Everything except the first item is optional. If your setup only does one thing, make it the shutdown, and make sure it is tested.

How software talks to the UPS

There are three physical paths, and the one your UPS supports determines which software can use it.

USB HID: the common denominator

The USB HID Power Device standard defines how a UPS reports charge, runtime and status in a vendor-neutral way. That is why plugging a typical UPS into a Windows PC makes a battery icon appear with no driver installed. It is also why NUT's usbhid-ups driver works across APC, CyberPower, Eaton, Tripp Lite and others. Vendors do add their own extensions, so some values (load percentage, input voltage, outlet controls) may be missing or reported differently depending on brand and firmware.

Serial: still around, easy to get wrong

Many business units keep a serial port, sometimes as an RJ45-shaped jack that is not Ethernet. Use the cable the manufacturer specifies. Some vendors use particular pins as "shut down now" signals, and the wrong cable can switch the UPS off when you connect it.

Network management cards and SNMP

A network management card (NMC) gives the UPS its own IP address, web interface, SNMP agent and email alerts. The UPS no longer depends on any one computer being alive to report its status, and shutdown agents on many servers can each subscribe to it. This is the cleanest architecture for a rack, at the cost of the card itself.

The three families of software

Built into the operating system

Windows treats a USB HID UPS as a battery and applies the "critical battery action" from Power Options; see Windows UPS shutdown. macOS shows UPS options in its energy settings when a compatible UPS is connected; see macOS UPS settings. NAS and hypervisor platforms embed an open source tool behind a settings page: Synology DSM and TrueNAS use NUT, Unraid uses apcupsd. Built-in support is the least work and usually enough for one machine.

Vendor software

APC's PowerChute and CyberPower's PowerPanel (covered in PowerChute and PowerPanel) expose brand-specific settings, read more detailed status, and include network shutdown agents designed for that vendor's management cards and virtualization platforms. They are the right choice when you need a vendor feature or vendor support. They lock you to one brand.

Open source

• Network UPS Tools (NUT) supports a very wide range of brands over USB, serial and SNMP, and has a clean client/server design. It is the engine behind many NAS and appliance UPS features.
• apcupsd is simpler, focused on APC units, and has a long track record. It sees little active development, but it is stable and still packaged by major distributions.

Which tool should you use?

One UPS, many machines

A small UPS often powers a NAS, a mini PC, a switch and a router. Only one of them can hold the USB cable. The pattern that works:

1. Pick the primary host. Choose the machine that stays up longest and is easiest to keep running: often the NAS or a small always-on Linux box. Connect the UPS USB cable to it.
2. Run a server on it. NUT in netserver mode (port 3493), apcupsd with NETSERVER on (port 3551), or a vendor agent in its "local" or server role.
3. Run clients on the others. Each client polls the primary for status and shuts itself down when the primary says the battery is critical or a forced shutdown is underway.
4. Order the shutdowns. Clients shut down first; the primary waits for them, then shuts itself down and tells the UPS to power off after a delay.
5. Keep the network path alive. The switch between the primary and its clients must also be on battery, or clients lose contact the instant power fails.

With a network management card, the card plays the primary role and every host is a client. The advantage is that the "server" is not a computer that might itself crash or be shutting down.

Picking shutdown triggers

Most tools offer three kinds of trigger. They can usually be combined, and whichever fires first wins.

• Time on battery ("shut down after 5 minutes on battery"). Predictable and independent of the UPS's estimates. Good when outages in your area are either very short or very long, because it skips the useless middle ground of draining the battery and crashing anyway.
• Remaining runtime ("shut down at 3 minutes left"). Adapts to load, but only as accurate as the UPS's runtime estimate, which drifts as batteries age until you recalibrate.
• Battery charge percentage ("shut down at 20%"). Simple, but charge percentage is a poor proxy for remaining time under a heavy load.
• The UPS's own low-battery signal. NUT and some vendor tools act on the UPS's internal low-battery flag. On some units this fires late, with well under a minute left, so many people raise the threshold or add a time-based trigger.

Testing that it actually works

A setup that has never been tested is a hope, not a plan. A safe test, done at a quiet time with backups current:

1. Confirm every host shows the UPS status as online (for example upsc for NUT or apcaccess for apcupsd).
2. Temporarily set a short trigger, such as "shut down after 1 minute on battery".
3. Simulate the outage by switching off the UPS's circuit breaker or unplugging its input cord. Watch the notifications arrive.
4. Confirm clients shut down first, then the primary, and note how long each took.
5. Restore power and check whether the UPS and machines come back as you expect.
6. Set the trigger back to its real value, informed by the times you measured.

Repeat after major OS upgrades and after replacing the UPS battery, since both can change behavior. The UPS maintenance checklist includes this as a periodic item.

Security notes for networked UPS software

UPS servers can do more than report status: a client with the right credentials can command a shutdown of every machine that trusts it. Keep the server listening only on a trusted LAN interface, use non-default passwords, block ports 3493 (NUT) and 3551 (apcupsd) at the router, and change the default credentials on any management card. Many NAS units ship with well-known default NUT credentials so that other devices can connect easily; that convenience is fine on an isolated home network and worth reconsidering anywhere else.