Nice introduction. I’m glad to see that the NixOS support for the Prometheus ecosystem is working for monitoring at an introductory level, and not just in production.
I would add that log management should not be considered as important as metrics when it comes to monitoring. Anything that needs to be operationalized for alerting can usually be formalized as a metric of some sort. Indeed, it seems that the article doesn’t list any concrete use cases for centralizing logging, and the most common use case I’ve seen in the wild is for log-based metrics, which should usually be replaced by service-specific exporters.
Thanks for this, I have Grafana setup already with InfluxDB and my weekend project is setting up Promethus + Loki!