I’m currently using InfluxDB + Telegraf + Grafana combination to monitor Linux systems and k3s pods. It’s basically same as Prometheus, but InfluxDB uses push model, which makes it easier to develop tools for collecting custom time series data.

For alerts and dashboards, I think Grafana is the simplest and most hassle free solution available at the moment.

I have a APC Back-UPS 1600VA. It powers two desktop PC/Server, a monitor, and router. So far, it gets the job done.

The biggest downside is; battery is not user replaceable, at least it’s not straight forward like the other models. If possible, prefer a UPS with the easy battery replacement option.

I’m actively using ollama with docker to run llama2:13b model. It’s generally works fine but heavy on resources as expected.

I personally use Traggo, but TimeTagger is also a great option: https://timetagger.app/articles/selfhost/