Show HN: I replaced Grafana+Prometheus with a Go binary and SSH for my VPSs

11 points

4 months ago

I do fullstack dev for work and side projects and recently moved everything to a couple VPSs on hetzner. Great setup, low cost, but one thing kept bugging me, how do I know if something is down without manually sshing in all the time? I tried the grafana + prometheus stack but the configuration time and seeing it use more resources than my actual apps was rough. Tried some smaller solutions too but nothing felt right. So I said screw it and built exactly what I wanted.

tori is a single go binary that runs on your server, reads host metrics from /proc and /sys, monitors containers via the docker socket (read-only), tails logs, and stores everything in sqlite with a 7 day retention. You define alert rules in a toml config and get notified via email or webhooks. The same binary runs on your local machine and connects over SSH via a TUI, no exposed ports, no HTTP server.

https://github.com/thobiasn/tori-cli

MIT licensed, happy to answer questions about the architecture or design choices :)