Victoriametrics

In Part 1, I walked through replacing Prometheus + node_exporter + cAdvisor with VictoriaMetrics + FluentBit on my homeserver. This is Part 2 — the actual numbers. All measurements are mean values over a 12-hour window, collected from the containers themselves via cAdvisor (before) and FluentBit docker metrics (after). They’re approximate, but they paint a clear enough picture. FluentBit vs Node Exporter + cAdvisor FluentBit replaced two containers — node_exporter and cAdvisor — with one. Here’s what that looks like in practice. ...

I’ve been running the same Prometheus monitoring stack on my homeservers for years. It worked fine — but it was more complex than it needed to be. node_exporter and cAdvisor are two separate containers doing one job: telling me what the machine and its containers are up to. And cAdvisor, it turns out, was the heaviest thing in my entire monitoring stack. I wanted something simpler and leaner. This is Part 1 — the motivation and the migration steps. Part 2 covers the actual before/after numbers. ...