cAdvisor
inari.cadvisor
Container resource monitor that discovers running containers automatically and exports per-container CPU, memory, filesystem and network statistics in Prometheus format and a built-in live view.
A Docker Compose file for cAdvisor alone, generated from the catalog and checked by the safety inspector. Sign in and use Kitchen Credits to activate it.
Uses the Medusa cart and payment-session path. This test control does not create an order or capture payment.
cAdvisor reads container and host statistics from the kernel's cgroup and namespace interfaces, so it collects usage for every running container without per-container configuration or an agent inside each one. It keeps a short in-memory history rendered as live charts and exposes the same data on a metrics endpoint for a scraper to retain long-term. Because it observes the runtime rather than the application, it supplies the resource picture — throttling, memory pressure, filesystem growth — that application-level metrics do not. A first deployment mounts the host paths it reads and confirms the container list appears.
You know it worked when
- The interface loads and lists the running containers.
- CPU and memory figures for a known container match its actual usage.
- The metrics endpoint returns Prometheus-formatted output.
- A scraper collects those metrics and they are queryable.
Known sharp edges
- It needs read access to several host paths and to the runtime's state directory, and a missing mount produces empty metrics rather than an error.
- The built-in interface keeps only a short window of history; anything longer requires scraping the metrics endpoint into a time-series store.
- Per-container metrics carry high label cardinality, and scraping every metric on a busy host puts real load on the storage behind it.
- The metrics endpoint is unauthenticated and discloses the full container inventory of the host.