← sources

Designing a self-hosted home alarm that works without a cloud: what would you want (and not trust)?

source: reddit (reddit:selfhosted)published 2026-09-28 09:26ingested 2026-09-29 02:40

retained text

I'm the lead developer of an open-source, Raspberry Pi based alarm system that I've been building for years. Before I decide what to work on next, I'd like some honest input from people who self-host critical stuff.

The current design: it replaces the mainboard of an old wired alarm, reuses existing sensors, siren and keypad, runs entirely locally, and talks to Home Assistant over MQTT. Alerts go out via email/SMS, and a GSM module covers internet outages.

A few questions I keep going back and forth on:

1. Would you expose an alarm's web UI to the internet (HTTPS + Let's Encrypt), or only behind a VPN? What would you expect from remote access?
2. What would stop you from trusting a DIY system with your home's alarm?
3. For Home Assistant users: is plain MQTT enough, or do you expect a native integration?
4. What's the failure mode you worry about most: power loss, network loss, SD card corruption, something else?

Happy to share the project link if anyone's interested, I didn't want this to be a plain ad.

document fetched 2026-09-29 00:30 · canonical https://reddit.com/r/selfhosted/comments/1ws9xdf/designing_a_selfhosted_home_alarm_that_works