⚠️ Deprecation notice

This preview integration of IoT Edge 1.x and Kubernetes will not be made generally available, and is no longer supported. The recommended way to run IoT Edge 1.x on Kubernetes is noted in the product's official docs


A Kubernetes cluster

A Kubernetes cluster that supports Custom Resource Definitions (CRD) is required for using Azure IoT Edge with Kubernetes. v1.12 or newer is recommended. Here are some options if you don't already have a cluster available:

For a managed cluster offering, consider Azure Kubernetes Service (AKS). You can stand one up using the AKS QuickStart instructions.

For a local development environment, you can use Minikube or k3d (recommended).

Persistent storage for the IoT Edge daemon (iotedged)

IoT Edge has a stateful architecture. Even if the workload modules are stateless, at a minimum, the context store for iotedged pod should be backed by non-ephemeral storage. Barring first-class support for native persistent volumes in your distribution, the local-path-provisioner provides a convenient way to utilize device local storage as a persistent store in a Kubernetes-native manner. For multi-node setup with node failure resilience, see the HA and scaling section.

For convenience, most tutorials that follow don't set up iotedged persistence. Persistence is not needed for getting things up and running but is highly recommended for demos, PoCs and, of course, production pilots. See the iotedged persistence tutorial for configuring state persistence for iotedged.

Helm 3

Helm is a package manager for Kubernetes which allows you to install applications, including IoT Edge, into your cluster. Please follow the Helm 3 install instructions.

The source code for Helm charts used by IoT Edge is available in the azure/iotedge repo


Kubectl is an essential tool for interacting with your cluster.

Visual Studio Code

Visual Studio Code with IoT Edge extension is recommended for working with IoT Edge's deployment manifests and submitting them to IoT Hub.

Visualization tools (optional)

Tools like Octant and K9s can help you understand the architecture and state of your cluster application.