// the find
stefanprodan/dockprom
Docker hosts and containers monitoring with Prometheus, Grafana, cAdvisor, NodeExporter and AlertManager
dockprom is a docker-compose stack that wires together Prometheus, Grafana, cAdvisor, Node Exporter, AlertManager, and Caddy into a single-command monitoring setup for Docker hosts. It ships pre-built Grafana dashboards covering host metrics, container metrics, and the monitoring stack itself. It's aimed at developers or small teams who want working observability on a single Docker host without configuring each component from scratch.
The pre-provisioned Grafana dashboards are the main value — you get useful, working visualizations immediately rather than staring at an empty Grafana instance. The alert rules are sensibly set up out of the box (high CPU load, memory, storage, container-down) covering the cases that actually matter for a single host. The Pushgateway inclusion is a good call for anyone with batch jobs that can't be scraped. The Caddy reverse proxy handles basic auth in front of Prometheus and AlertManager, which are otherwise completely open by default.
The repo is essentially frozen at 2019-era component versions — Grafana 7.2, alert rules using `aufs` as the default fstype when most modern Linux systems use `overlay2`, which means the storage alert fires on nothing out of the box. Container-specific alerts (the Jenkins examples) are hardcoded in the alert rules file rather than templated, so you have to edit YAML every time you add a service. There's no TLS termination for the Grafana endpoint — Caddy protects Prometheus and AlertManager but Grafana sits exposed on port 3000. For anything beyond a single host or a handful of containers this falls apart immediately; there's no service discovery, no multi-host support, and no guidance on scaling the Prometheus retention.