Pod CleanUp

Introduction

This is a Helm chart for Kubernetes that automates the cleanup of failed pods across all namespaces. It uses a CronJob to periodically execute the kubectl delete pods command to remove pods that are in the Failed state.

Prerequisites

• Kubernetes 1.25+
• Helm 3.0+

Installing on your cluster

The default values for this chart are defined in the values.yaml file. You can customize the settings by passing your own values during installation:

helm install pod-cleanup oci://ghcr.io/cubos/charts/pod-cleanup \
  --set podCleanup.schedule="0 2 * * *" \
  --version <version>

Configurable Values

• podCleanup.cleanupContainer.image.repository: Repository of the cleanup container image (default: bitnami/kubectl).
• podCleanup.cleanupContainer.image.tag: Tag of the cleanup container image (default: latest).
• podCleanup.schedule: Cron schedule for the CronJob (default: 0 3 * * *).
• podCleanup.serviceAccount.annotations: Annotations for the service account.

Usage

This chart creates a CronJob that executes the command kubectl delete pods --field-selector status.phase==Failed --all-namespaces according to the configured cron schedule. Ensure that the service account has the necessary permissions to list and delete pods.

Contributing

PRs are welcome! Please make sure to check ESLint and Prettier before submitting a PR.

By submitting a PR, you agree to license your work under the GNU Lesser General Public License v3.0.