Skip to content

howto-prepare-cluster.md

Latest commit

 

History

History
346 lines (222 loc) · 21.1 KB

howto-prepare-cluster.md

File metadata and controls

346 lines (222 loc) · 21.1 KB
title Prepare your Kubernetes cluster
description Prepare an Azure Arc-enabled Kubernetes cluster before you deploy Azure IoT Operations. This article includes guidance for both Ubuntu and Windows machines.
author dominicbetts
ms.author dobett
ms.topic how-to
ms.custom ignite-2023, devx-track-azurecli
ms.date 10/23/2024

Prepare your Azure Arc-enabled Kubernetes cluster

An Azure Arc-enabled Kubernetes cluster is a prerequisite for deploying Azure IoT Operations. This article describes how to prepare a cluster before you deploy Azure IoT Operations. This article includes guidance for Ubuntu, Windows, Azure Local, and vSphere Kubernetes Service (VKS), formerly Tanzu Kubernetes Grid Service.

If you want to deploy Azure IoT Operations quickly and run a sample workload in a test environment, see the Quickstart: Run Azure IoT Operations in GitHub Codespaces with K3s.

Prerequisites

For multi-node deployments, K3s for deployments on Ubuntu, AKS deployments on Azure Local, and VKS. Azure Kubernetes Service (AKS) Edge Essentials for Windows only supports single-node deployments.

Ubuntu

To prepare an Azure Arc-enabled Kubernetes cluster, you need:

[!INCLUDE Cluster prerequisites for Ubuntu and VKS]

AKS Edge Essentials

To prepare an Azure Arc-enabled Kubernetes cluster, you need:

AKS on Azure Local

To prepare an Azure Arc-enabled Kubernetes cluster, you need:

VKS

To prepare a VKS cluster, you need:

[!INCLUDE Cluster prerequisites for Ubuntu and VKS]

Create and Arc-enable a cluster

This section provides steps to create clusters in validated environments on Linux and Windows.

Ubuntu

To prepare a K3s Kubernetes cluster on Ubuntu:

  1. Create a single-node or multi-node K3s cluster. For examples, see the K3s quick-start guide or K3s related projects.

  2. Check to see that kubectl was installed as part of K3s. If not, follow the instructions to Install kubectl on Linux.

    kubectl version --client

  3. Follow the instructions to Install Helm.

  4. Create a K3s configuration yaml file in .kube/config:

    mkdir ~/.kube
sudo KUBECONFIG=~/.kube/config:/etc/rancher/k3s/k3s.yaml kubectl config view --flatten > ~/.kube/merged
mv ~/.kube/merged ~/.kube/config
chmod  0600 ~/.kube/config
export KUBECONFIG=~/.kube/config
#switch to k3s context
kubectl config use-context default
sudo chmod 644 /etc/rancher/k3s/k3s.yaml

  5. Run the following command to increase the user watch/instance limits.

    echo fs.inotify.max_user_instances=8192 | sudo tee -a /etc/sysctl.conf
echo fs.inotify.max_user_watches=524288 | sudo tee -a /etc/sysctl.conf

sudo sysctl -p

  6. For better performance, increase the file descriptor limit:

    echo fs.file-max = 100000 | sudo tee -a /etc/sysctl.conf

sudo sysctl -p

Arc-enable your cluster

Connect your cluster to Azure Arc so that it can be managed remotely.

  1. From a machine that has kubectl access to your cluster, sign into Azure CLI with your Microsoft Entra user account that has the required role(s) for the Azure subscription:

    az login

    If at any point you get an error that says Your device is required to be managed to access your resource, run az login again and make sure that you sign in interactively with a browser.

  2. After you sign in, the Azure CLI displays all of your subscriptions and indicates your default subscription with an asterisk *. To continue with your default subscription, select Enter. Otherwise, type the number of the Azure subscription that you want to use.

  3. Register the required resource providers in your subscription.

    [!NOTE] This step only needs to be run once per subscription. To register resource providers, you need permission to do the /register/action operation, which is included in subscription Contributor and Owner roles. For more information, see Azure resource providers and types.

    az provider register -n "Microsoft.ExtendedLocation"
az provider register -n "Microsoft.Kubernetes"
az provider register -n "Microsoft.KubernetesConfiguration"
az provider register -n "Microsoft.IoTOperations"
az provider register -n "Microsoft.DeviceRegistry"
az provider register -n "Microsoft.SecretSyncController"

  4. Use the az connectedk8s connect command to Arc-enable your Kubernetes cluster and manage it as part of your Azure resource group.

    az connectedk8s connect --name <CLUSTER_NAME> -l <REGION> --resource-group <RESOURCE_GROUP> --subscription <SUBSCRIPTION_ID> --enable-oidc-issuer --enable-workload-identity --disable-auto-upgrade

    To prevent unplanned updates to Azure Arc and the system Arc extensions that Azure IoT Operations uses as dependencies, this command disables autoupgrade. Instead, manually upgrade agents as needed.

    [!IMPORTANT] If your environment uses a proxy server or Azure Arc Gateway, modify the az connectedk8s connect command with your proxy information:

    1. Follow the instructions in either Connect using an outbound proxy server or Onboard Kubernetes clusters to Azure Arc with Azure Arc Gateway.
    2. Add 169.254.169.254 to the --proxy-skip-range parameter of the az connectedk8s connect command. Azure Device Registry uses this local endpoint to get access tokens for authorization.

    Azure IoT Operations doesn't support proxy servers that require a trusted certificate.

  5. Get the cluster's issuer URL.

    az connectedk8s show --resource-group <RESOURCE_GROUP> --name <CLUSTER_NAME> --query oidcIssuerProfile.issuerUrl --output tsv

    Save the output of this command to use in the next steps.

  6. Create a k3s config file.

    sudo nano /etc/rancher/k3s/config.yaml

  7. Add the following content to the config.yaml file, replacing the <SERVICE_ACCOUNT_ISSUER> placeholder with your cluster's issuer URL.

    kube-apiserver-arg:
 - service-account-issuer=<SERVICE_ACCOUNT_ISSUER>
 - service-account-max-token-expiration=24h

  8. Save the file and exit the nano editor.

  9. Prepare for enabling the Azure Arc service, custom location, on your Arc cluster by getting the custom location object ID and saving it as the environment variable, OBJECT_ID. You must be logged into Azure CLI with a Microsoft Entra user account to successfully run the command, not a service principal. Run the following command exactly as written, without changing the GUID value.

    export OBJECT_ID=$(az ad sp show --id bc313c14-388c-4e7d-a58e-70017303ee3b --query id -o tsv)

    [!NOTE] If you receive the error: "Unable to fetch oid of 'custom-locations' app. Proceeding without enabling the feature. Insufficient privileges to complete the operation," then your service principal might lack the necessary permissions to retrieve the object ID of the custom location. Log into Azure CLI with a Microsoft Entra user account that meets the prerequisites. For more information, see Create and manage custom locations.

  10. Use the az connectedk8s enable-features command to enable the custom location feature on your Arc cluster. This command uses the OBJECT_ID environment variable saved from the previous step to set the value for the custom-locations-oid parameter. Run this command on the machine where you deployed the Kubernetes cluster:

    az connectedk8s enable-features -n <CLUSTER_NAME> -g <RESOURCE_GROUP> --custom-locations-oid $OBJECT_ID --features cluster-connect custom-locations

  11. Restart K3s.

    systemctl restart k3s

Configure multi-node clusters for Azure Container Storage enabled by Azure Arc

Features such as data flow local storage endpoints and the media connector optionally use Azure Container Storage enabled by Azure Arc (ACSA) to synchronize local data to the cloud. ACSA is not installed as part of Azure IoT Operations, so you must install it separately.

On multi-node Ubuntu clusters with at least three nodes, you have the option to enable fault tolerance for ACSA storage. To enable fault tolerance during deployment, follow the steps in Prepare Linux for Edge Volumes using a multi-node Ubuntu cluster to configure your cluster.

If you're running your cluster on a Kubernetes distribution other than k3s, review the guidance to Prepare Linux with other platforms.

AKS Edge Essentials

Azure Kubernetes Service Edge Essentials is an on-premises Kubernetes implementation of Azure Kubernetes Service (AKS) that automates running containerized applications at scale. AKS Edge Essentials includes a Microsoft-supported Kubernetes platform that includes a lightweight Kubernetes distribution with a small footprint and simple installation experience that supports PC-class or "light" edge hardware.

The AksEdgeQuickStartForAio.ps1 script automates the process of creating and connecting a cluster, and is the recommended path for deploying Azure IoT Operations on AKS Edge Essentials.

For instructions on running the script, see Configure an AKS Edge Essentials cluster for Azure IoT Operations.

AKS on Azure Local

By default, a Kubernetes cluster is created with a node pool that can run Linux containers. If you add more node pools after creation, make sure the OS is set to Linux. Azure IoT Operations doesn't support deployment to Windows nodes.

Then, once you have an Azure Arc-enabled Kubernetes cluster, you can deploy Azure IoT Operations.

VKS

For guidance on setting up a VKS cluster, see Deploying VKS Service Clusters.

Important

When you initialize your Azure IoT Operations instance on a VKS cluster with its pod security admission controller running in restricted mode you must include the --cm-config global.telemetry.logs.enabled=false flag when you run az iot ops init.

Arc-enable your cluster

Connect your cluster to Azure Arc so that it can be managed remotely.

  1. On the machine where you deployed the Kubernetes cluster, sign into Azure CLI with your Microsoft Entra user account that has the required role(s) for the Azure subscription:

    az login

  2. After you sign in, the Azure CLI displays all of your subscriptions and indicates your default subscription with an asterisk *. To continue with your default subscription, select Enter. Otherwise, type the number of the Azure subscription that you want to use.

  3. Register the required resource providers in your subscription.

    [!NOTE] This step only needs to be run once per subscription. To register resource providers, you need permission to do the /register/action operation, which is included in subscription Contributor and Owner roles. For more information, see Azure resource providers and types.

    az provider register -n "Microsoft.ExtendedLocation"
az provider register -n "Microsoft.Kubernetes"
az provider register -n "Microsoft.KubernetesConfiguration"
az provider register -n "Microsoft.IoTOperations"
az provider register -n "Microsoft.DeviceRegistry"
az provider register -n "Microsoft.SecretSyncController"

  4. Use the az connectedk8s connect command to Arc-enable your Kubernetes cluster and manage it as part of your Azure resource group.

    az connectedk8s connect --name <CLUSTER_NAME> -l <REGION> --resource-group <RESOURCE_GROUP> --subscription <SUBSCRIPTION_ID> --enable-oidc-issuer --enable-workload-identity --disable-auto-upgrade

    To prevent unplanned updates to Azure Arc and the system Arc extensions that Azure IoT Operations uses as dependencies, this command disables autoupgrade. Instead, manually upgrade agents as needed.

    [!IMPORTANT] If your environment uses a proxy server or Azure Arc Gateway, modify the az connectedk8s connect command with your proxy information:

    1. Follow the instructions in either Connect using an outbound proxy server or Onboard Kubernetes clusters to Azure Arc with Azure Arc Gateway.
    2. Add 169.254.169.254 to the --proxy-skip-range parameter of the az connectedk8s connect command. Azure Device Registry uses this local endpoint to get access tokens for authorization.

    Azure IoT Operations doesn't support proxy servers that require a trusted certificate.

  5. Get the cluster's issuer URL.

    az connectedk8s show --resource-group <RESOURCE_GROUP> --name <CLUSTER_NAME> --query oidcIssuerProfile.issuerUrl --output tsv

Save the output of this command to use in the next steps.

  1. Connect to the VKS cluster. Edit the custom resource for the workload cluster with the issuer URL from the previous step.

    kubectl edit cluster <CLUSTER_NAME>

  2. Add the following content to the config.yaml file, replacing the <OIDC_ISSUER_URL> placeholder with your cluster's issuer URL.

    [!NOTE] The URL should be copied exactly as printed by the prior command, including any characters such as /.

    - name: apiServerExtraArgs
  value: {"service-account-issuer":"<OIDC_ISSUER_URL>"}

  3. Prepare for enabling the Azure Arc service, custom location, on your Arc cluster by getting the custom location object ID and saving it as the environment variable, OBJECT_ID. You must be logged into Azure CLI with a Microsoft Entra user account to successfully run the command, not a service principal. Run the following command exactly as written, without changing the GUID value.

    export OBJECT_ID=$(az ad sp show --id bc313c14-388c-4e7d-a58e-70017303ee3b --query id -o tsv)

    [!NOTE] If you receive the error: "Unable to fetch oid of 'custom-locations' app. Proceeding without enabling the feature. Insufficient privileges to complete the operation," then your service principal might lack the necessary permissions to retrieve the object ID of the custom location. Log into Azure CLI with a Microsoft Entra user account that meets the prerequisites. For more information, see Create and manage custom locations.

  4. Use the az connectedk8s enable-features command to enable the custom location feature on your Arc cluster. This command uses the OBJECT_ID environment variable saved from the previous step to set the value for the custom-locations-oid parameter. Run this command on the machine where you deployed the Kubernetes cluster:

    az connectedk8s enable-features -n <CLUSTER_NAME> -g <RESOURCE_GROUP> --custom-locations-oid $OBJECT_ID --features cluster-connect custom-locations

Advanced configuration

At this point, when you have an Azure Arc-enabled Kubernetes cluster but before you deploy Azure IoT Operations to it, you might want to configure your cluster for advanced scenarios.

Next steps

Now that you have an Azure Arc-enabled Kubernetes cluster, you can choose to deploy Azure IoT Operation with test settings or with production settings.

  • Test deployment: Recommended for quick evaluation and prototyping before deploying in production. Test deployment isn't suitable for production, it lacks observability and hardened security.
  • Production deployment: Recommended for production-ready workloads. Production deployment is suitable for real-world IoT deployments with compliance and security needs.