Overview

Kind (Kubernetes IN Docker) is a tool for running local Kubernetes clusters using Docker containers as cluster nodes. Developed by the Kubernetes SIG Testing, kind was designed primarily for testing Kubernetes itself but has become widely adopted for local development and CI workflows.

Kind bootstraps each “node” with kubeadm and runs systemd inside Docker containers. With over 15,000 GitHub stars and 4,800+ commits, kind is the fastest way to spin up a Kubernetes cluster for testing — a cluster can be ready in under 30 seconds, making it ideal for CI pipelines that need ephemeral, disposable clusters.

Key Features

• Lightning-fast cluster creation — Create a multi-node Kubernetes cluster in seconds using Docker containers as nodes; kind create cluster is all you need
• Multi-node clusters — Define control-plane and worker nodes with a simple YAML configuration file for testing node-level scenarios
• Built-in image building — Build Kubernetes node images from source with kind build node-image for testing custom Kubernetes builds
• Kubeadm-based bootstrap — Uses kubeadm for cluster initialization, ensuring compatibility with upstream Kubernetes
• Port mapping — Map container ports to host ports for accessing applications running in the cluster
• Disposable by design — Clusters are ephemeral containers; delete with kind delete cluster for a complete cleanup with no leftover state
• Offline and air-gapped support — Pre-load container images for use in restricted network environments
• kubetest integration — Native support in the Kubernetes test framework (kubetest) for running conformance and end-to-end tests
• Custom node images — Build and use custom node images with specific Kubernetes versions, patches, or configurations
• Export cluster logs — Export logs from all cluster nodes with kind export logs for debugging test failures

Why Use It

Kind is purpose-built for speed and ephemerality. Unlike full VM-based solutions, kind runs nodes as Docker containers, eliminating virtualization overhead and enabling cluster creation in under 30 seconds. This makes it the ideal choice for CI pipelines where every second counts and clusters are created and destroyed hundreds of times per day. It’s also the tool Kubernetes itself uses for its own end-to-end testing — if it’s good enough for the project’s own CI, it’s good enough for yours.

Use Cases

• CI/CD pipelines — Spin up ephemeral Kubernetes clusters in CI for integration and end-to-end testing, then destroy them
• Kubernetes development — Test new Kubernetes features, custom controllers, and operators against specific API versions
• Conformance testing — Run Kubernetes conformance tests (Sonobuoy) against custom cluster configurations
• Local development — Quickly test Kubernetes manifests, Helm charts, and Kustomize overlays before deploying to production
• Kubernetes SIG Testing — Official tool for testing Kubernetes itself; used in the project’s own CI pipeline
• Multi-version testing — Test workloads against multiple Kubernetes versions simultaneously using different kind clusters

Platforms

Linux · macOS · Windows

Licence

Apache License 2.0

Website

kind.sigs.k8s.io · GitHub