Eraser
Eraser is a Kubernetes-native container image cleanup system designed to automatically detect and remove unused images across cluster nodes to reduce disk usage and security exposure (container lifecycle management, Kubernetes operations).
- Automated removal of unused container images from Kubernetes nodes (container lifecycle management).
- Policy-driven cleanup based on configurable rules and scheduling (policy-based automation).
- Integration with Kubernetes primitives such as DaemonSets, Jobs, and CRDs for cluster-wide operation (Kubernetes operations).
- Support for multiple registries and image sources as defined by cluster workloads (container registry interoperability).
- Reduction of node disk consumption and stale image surface area in multi-tenant or long-lived clusters (infrastructure optimization, security hygiene).
More About Eraser
Eraser is an open-source project under the Cloud Native Computing Foundation (CNCF) ecosystem that focuses on automated cleanup of unused container images on Kubernetes clusters (container lifecycle management). The project addresses the accumulation of stale images on worker nodes, which can increase disk consumption and leave outdated, unused artifacts present on production infrastructure. By automating removal of images that are no longer referenced by running workloads, Eraser supports more predictable node capacity planning and reduces the maintenance overhead associated with manual cleanup.
The core function of Eraser is to scan nodes in a Kubernetes cluster and identify container images that are safe to delete based on configurable rules and policies (policy-based automation). It typically runs as Kubernetes-native components such as DaemonSets and Jobs (Kubernetes operations), enabling cluster administrators to deploy image cleanup logic using standard manifest-based workflows. Through these primitives, Eraser can operate across all nodes in a cluster, enforcing cleanup behavior in a consistent manner and aligning with established DevOps and GitOps practices for infrastructure management.
Configuration in Eraser centers on defining which images should be retained and which can be removed, often based on factors such as whether images are currently in use by pods, allowed registries, or other policy criteria (governance and compliance). Administrators can schedule cleanup operations at intervals to run during maintenance windows or at times that align with workload patterns, reducing contention for disk and I/O resources. The project also provides status and reporting through Kubernetes resources, allowing operators to observe which images were deleted and the state of cleanup tasks.
In enterprise environments, Eraser is positioned as a cluster hygiene and resource management tool for Kubernetes platforms (platform operations). It can be integrated into managed or self-hosted Kubernetes clusters to prevent disk exhaustion on nodes, especially in environments with frequent deployments, image rebuilds, or ephemeral workloads. By reducing the footprint of unused images, Eraser also narrows the set of artifacts present on production nodes, which can support security practices that focus on minimizing exposed software components (security hygiene).
From an architectural perspective, Eraser aligns with Kubernetes-native patterns, using Custom Resource Definitions (CRDs) and controller logic to express cleanup policies and orchestrate removal workflows (cloud-native automation). This approach allows Eraser to interoperate with other Kubernetes tooling and automation pipelines that rely on declarative configuration and reconciliation loops. In a technical directory or taxonomy, Eraser fits within categories such as container lifecycle management, Kubernetes cluster operations, and infrastructure optimization tools that manage node-level resources and artifacts.