Many home servers should target at least 10โ15 minutes of usable runtime at the measured load, but the correct number is outage delay plus guest and storage shutdown time plus battery reserve. Size from watts and a runtime curve, not the UPS VA label alone.
If the goal is graceful shutdown rather than continued operation, longer runtime is useful only when it creates a safer margin. Aging batteries, high load, cold starts, and network dependencies can shrink the time available. This distinction sets the measurement method, safety margin, and stop condition. This distinction sets the measurement method, safety margin, and stop condition. This distinction sets the measurement method, safety margin, and stop condition.
Measure the complete shutdown chain
Time the delay before shutdown begins, the slowest VM or container stop, database checkpoints, storage unmounts, and the host power-off. Include the UPS monitor and network switch if another machine depends on them.
Measure connected watts during a representative heavy workload. A shutdown that starts during a scrub, transcode, or backup may take longer and draw more power than an idle test.
Add reserve after the host is off so battery voltage does not collapse at the final write. A practical plan often keeps 25โ50% of tested runtime unused at the shutdown point.
Choose thresholds from battery condition
Use the manufacturer runtime chart for the exact model and load, then validate with a controlled test. Linear calculations from VA or battery watt-hours can be misleading because inverter efficiency and discharge behavior change with load.
Retest after battery replacement and at least periodically as the pack ages. A self-test that reports healthy does not prove the original runtime remains available.
Use the table below to decide whether the current UPS has enough margin.
| Observed state | Verdict | Next action |
|---|---|---|
| Shutdown plus 25โ50% reserve fits | Adequate | Retest as battery ages |
| Host finishes with little reserve | Marginal | Trigger earlier or increase capacity |
| UPS cuts output before host stops | Unsafe | Reduce load or replace UPS |
Configure communication and order
Connect the UPS by USB, SNMP, or a supported network monitor and make shutdown automatic. Shut down applications and guests before the hypervisor, and keep the switch powered until network clients receive the event.
Avoid setting the low-battery threshold so late that the server races the UPS cutoff. Prefer an early, deterministic shutdown with capacity left for filesystem sync.
ZimaSpaceโs UPS shutdown-chain test covers host and guest timing.
rHomelabโs UPS sizing and NUT walkthrough explains why watts, runtime, and automatic shutdown must be evaluated together.
Prove runtime with a controlled outage
With current backups verified, remove mains input from the UPS while leaving its outputs connected. Record each event and restore mains before reserve is exhausted on the first test.
Repeat at the intended shutdown threshold and confirm every guest stops, filesystems unmount, and the host powers off before the UPS reaches its cutoff. Inspect logs after restart.
Proceed when worst-case shutdown plus reserve fits inside measured aged-battery runtime. Increase runtime or trigger earlier if margin is thin; stop testing if the UPS overheats, overloads, or battery voltage falls unpredictably.
Support & Tips
More to Read

What to Check Before Loading a Community Model in Transformers
Use the Transformers Jinja sandbox update to review model provenance, chat templates, remote code, serialized weights, and isolated first runs.

Run Mirage Locally: OCR and Object Detection from One Image
Try Cactus Computeโs small local vision model, validate OCR and object extraction on real inputs, and understand memory, language, and privacy limits.

How to Schedule a Coding Agent in Open WebUI on a Home Server
Set up a report-only coding agent automation in Open WebUI, select its terminal and workspace, and verify permissions, logs, and failure recovery.



