cadvisor

What is cadvisor?

The cAdvisor (Container Advisor) image provides a lightweight tool for analyzing the performance and resource usage of running containers. It collects, aggregates, and exports metrics such as CPU, memory, filesystem, and network usage for each container, making it a foundational component of container observability stacks.

Originally developed by Google, cAdvisor is often run as a daemonset in Kubernetes clusters to provide node-level container insights. It exposes metrics over a REST API and in Prometheus format, allowing integration into monitoring pipelines.

Because of its low overhead and native container support, cAdvisor is widely used to give operators and developers visibility into container health and efficiency.

How to use this image

The cAdvisor image can be run as a standalone container or deployed in a Kubernetes cluster.

Examples:

<code># Run cAdvisor standalone and expose metrics on port 8080
docker run \
 --volume=/:/rootfs:ro \
 --volume=/var/run:/var/run:rw \
 --volume=/sys:/sys:ro \
 --volume=/var/lib/docker/:/var/lib/docker:ro \
 --publish=8080:8080 \
 --detach \
 gcr.io/cadvisor/cadvisor:latest</code>

<code># Access the cAdvisor web UI
http://localhost:8080</code>

<code># Scrape metrics with Prometheus
prometheus.yml:
 scrape_configs:
   - job_name: 'cadvisor'
     static_configs:
       - targets: ['localhost:8080']</code>

In Kubernetes, cAdvisor functionality is already included in the kubelet binary, but running the image separately can provide richer metrics or custom configurations.

Image variants

The cAdvisor image is typically available in a few main variants:

cadvisor:<version>

Version-pinned images (e.g., cadvisor:v0.49.1) are recommended for production to ensure consistent behavior.

cadvisor:latest

Tracks the most recent stable release. Good for testing, but not ideal for production due to version drift.

cadvisor:canary

Built from the latest development commits. Useful for contributors or early adopters testing unreleased features.