Thousands of cAdvisor Docker containers created until disk runs out of space

Google Compute Engine container-optimized VM image has a built-in cAdvisor container that gets started automatically.

In the last three days, an infinite loop started to create ~15000 containers of the cAdvisor image until the disk ran out of space.

  • Docker - Telnet from container to host
  • How-to: Dockerfile layer(s) reuse / reference
  • AWS ECS - how to log to cloudwatch from ECS container?
  • How does rancher schedule services to specific hosts?
  • Docker Service Separation
  • Docker-Compose: how to mapping volume from host into container?
  • The versions are:

    Linux docker-04 3.16-0.bpo.2-amd64 #1 SMP Debian 3.16.3-2~bpo70+1 (2014-09-21) x86_64

    Docker version 1.2.0, build fa7b24f

    How can this bug be fixed?

    For a temporary fix, I’ve deleted all ~15000 containers using the following script:

    docker ps -a | grep ‘days ago’ | awk ‘{print $1}’ | xargs –no-run-if-empty docker rm

  • ZeroMQ fails to .bind() on Docker on [] - address already in use. Why?
  • Using Docker Compose to connect to mysql running in another container
  • Map docker container to regular dev domain name
  • JVM Memory Allocation In Docker (LXC) Container
  • Permission required to run java under non-root user in Google Container Engine
  • How to know if my program is completely started inside my docker with compose
  • One Solution collect form web for “Thousands of cAdvisor Docker containers created until disk runs out of space”

    There was a bug in the latest cAdvisor release (0.7.0). It has been fixed since and a patch release was made (0.7.1). GCE container image should pick it up and release a new build soon.

    If you see any issues, please report them via the GitHub issue tracker.

    Docker will be the best open platform for developers and sysadmins to build, ship, and run distributed applications.