# What should shut down first, a NAS or the machine writing to it? — retained record

> Retained reddit record from reddit:homelab.

Canonical: https://highsignal.app/data/records/fb1e3923755c77be

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:homelab)  published 2026-09-23 14:59 ingested 2026-09-24 02:39

[original source ↗](https://www.reddit.com/r/homelab/comments/1wo8dm2/what_should_shut_down_first_a_nas_or_the_machine/)

## retained text



```
My small lab in Germany has a two-drive NAS for photo backups and a Linux mini PC that imports new folders and copies them to the NAS. I started making an outage list while considering a Jackery SolarVault 3 Pro for the wider household. When I put the NAS down as something I wanted to keep running, I noticed that a copy job was still active. Keeping the NAS powered for a while was only half an answer; the two computers also needed an orderly way to stop.

For the NAS, I am also planning a small NUT-compatible UPS to provide the shutdown signal. The awkward case is a copy already in progress when that UPS reaches its shutdown threshold. My current sequence is for the mini PC to stop new jobs, allow the active job to finish within a hard limit or terminate it cleanly, and only then let the NAS shut down. The network switch also has to remain up long enough to carry the notifications. Waiting indefinitely for a stuck copy would defeat the sequence.

The NUT documentation says a primary can wait for secondaries, subject to a timeout. I still need to establish which host should connect to the UPS and expose its status through upsd, whether both machines can run upsmon, and which one should be primary.

How do people here handle a write client that is still active when the NAS has to stop, especially when there is too little runtime left to finish the current job?
```



document fetched 2026-09-24 00:39 · canonical https://reddit.com/r/homelab/comments/1wo8dm2/what_should_shut_down_first_a_nas_or_the_machine
