What it does
cAdvisor is a host-side daemon for inspecting container resource use and performance. It gathers container and machine statistics, keeps historical samples, and makes them available through a web interface, storage integrations, and a versioned REST API. The project has native Docker support and aims to work with other container types through its container abstraction. It is a building block for container visibility, not a complete hosted monitoring and incident-management service.
Read container-level resource data
The upstream quick start runs cAdvisor on a machine whose containers need observing. Its provided Docker example mounts host filesystem and runtime paths, starts the cAdvisor image, and exposes the web interface on port 8080. Once deployed, an operator can inspect per-container resource usage and performance history instead of relying only on host-wide CPU or memory totals. The example is documented, not executed here.
cAdvisor also exposes raw and processed statistics through a versioned remote REST API, and its repository points to storage plugins for exporting collected data. That makes it useful as an input to an existing metrics or visualization stack. For example, an operator investigating a resource-use change could compare historical statistics for individual containers with machine-wide data, then query or export those measurements for another dashboard.
Check the host access before deploying
This is a daemon with visibility into the host, not a remote SaaS monitor. The official Docker quick-start uses --privileged and mounts paths such as /sys, /var/run, and the root filesystem so cAdvisor can inspect runtime and machine data. Review the required mount and permission model for the chosen installation instead of copying that example blindly into a restricted environment. The source repository uses the Apache License 2.0.
Strengths
- Provides historical resource and performance statistics per container as well as machine-wide data.
- The upstream quick start offers a container image and a built-in web interface for a single host.
- A versioned REST API and storage plugins make the collected statistics available to other systems.
Watch out for
- The documented Docker quick-start requires broad host visibility, including privileged operation and host-path mounts; deployment permissions need review.
- It collects and exposes measurements but leaves dashboards, alert routing, and incident response to the surrounding monitoring stack.
Best for
- Inspecting resource use and network statistics for containers on a host
- Feeding container statistics into an existing metrics backend or dashboard
- Operators who need a self-hosted view of container and machine resource history