/dcc

Dangling Container Checker

Primary LanguageGoMIT LicenseMIT

Dangling Container Checker

The purpose of Dangling Container Checker is to watch over Docker and Kubernetes to ensure containers are not being forgotten by Kubernetes. When a dangling container is found, DCC can either report it to Kubernetes Events or kill the container.

How to Deploy

Deploy this utility as a DaemonSet to monitor your entire Kubernetes cluster.

How to Configure

config.yaml Settings

Optionally, along side the DaemonSet, deploy a ConfigMap with config.yaml as the data
and mount it in the DCC as a volumeMount in /config.

Timing

The interval between checks and the stop grace period timeout are both configurable with check_interval and stop_timeout.

timing:
  check_interval: 60
  stop_timeout: 30
Whitelisting Image Names

Some containers are not kept in Kubernetes records, such as the "pause" container. By adding its image name to the whitelist, ContainerChk will ignore these containers in Docker. It's also feasible that you may not want to run ContainerChk as a Kubernetes workload, so "containerchk" is also added to the image whitelist by default.

whitelist:
  images:
    - gcr.io/google_containers/pause-amd64
    - containerchk