Edit

Deploy to Kubernetes

  • Article

Azure DevOps Services | Azure DevOps Server 2022

You can use Azure Pipelines to deploy to Azure Kubernetes Service and Kubernetes clusters offered by other cloud providers. Azure Pipelines has two tasks for working with Kubernetes:

  • KubernetesManifest task: bake and deploy manifests to Kubernetes clusters with Helm, Kompose, or Kustomize
  • Kubectl task: deploy, configure, and update a Kubernetes cluster in Azure Container Service by running kubectl commands

If you're using Azure Kubernetes Service with either task, the Azure Resource Manager service connection type is the best way to connect to a private cluster, or a cluster that has local accounts disabled.

For added deployment traceability, use a Kubernetes resource in environments with a Kubernetes task.

To get started with Azure Pipelines and Azure Kubernetes service, see Build and deploy to Azure Kubernetes Service with Azure Pipelines. To get started with Azure Pipelines, Kubernetes, and the canary deployment strategy specifically, see Use a canary deployment strategy for Kubernetes deployments with Azure Pipelines.

KubernetesManifest task

The KubernetesManifest task checks for object stability before marking a task as success/failure. The task can also perform artifact substitution, add pipeline traceability-related annotations, simplify creation and referencing of imagePullSecrets, bake manifests, and aid in deployment strategy roll outs.

Note

While YAML-based pipeline support triggers on a single Git repository, if you need a trigger for a manifest file stored in another Git repository or if triggers are needed for Azure Container Registry or Docker Hub, you should use a classic pipeline instead of a YAML-based pipeline.

You can use the bake action in the Kubernetes manifest task to bake templates into Kubernetes manifest files. The action lets you use tools such as Helm, Kustomize, and Kompose. The bake action of the Kubernetes manifest task provides visibility into the transformation between input templates and the end manifest files that are used in deployments. You can consume baked manifest files downstream (in tasks) as inputs for the deploy action of the Kubernetes manifest task.

You can target Kubernetes resources that are part of environments with deployment jobs. Using environments and resources deployment gives you access to better pipeline traceability so that you can diagnose deployment issues. You can also deploy to Kubernetes clusters with regular jobs without the same health features.

The following YAML code is an example of baking manifest files from Helm charts

steps:
- task: KubernetesManifest@0
  name: bake
  displayName: Bake K8s manifests from Helm chart
  inputs:
    action: bake
    helmChart: charts/sample
    overrides: 'image.repository:nginx'

- task: KubernetesManifest@0
  displayName: Deploy K8s manifests
  inputs:
    kubernetesServiceConnection: someK8sSC
    namespace: default
    manifests: $(bake.manifestsBundle)
    containers: |
      nginx: 1.7.9

Kubectl task

As an alternative to the KubernetesManifest KubernetesManifest task, you can use the Kubectl task to deploy, configure, and update a Kubernetes cluster in Azure Container Service by running kubectl commands.

The following example shows how a service connection is used to refer to the Kubernetes cluster.

- task: Kubernetes@1
  displayName: kubectl apply
  inputs:
    connectionType: Kubernetes Service Connection
    kubernetesServiceEndpoint: Contoso

Script task

You can also use kubectl with a script task.

The following example uses a script to run kubectl.

- script: |
    kubectl apply -f manifest.yml

Kubernetes deployment strategies

The Kubernetes manifest task currently supports canary deployment strategy. Use canary deployment strategy to partially deploy new changes so that the new changes coexist with the current deployments before a full rollout.

For more information on canary deployments with pipelines, see Use a canary deployment strategy for Kubernetes deployments with Azure Pipelines.

Multicloud Kubernetes deployments

Kubernetes runs the same way on all cloud providers. Azure Pipelines can be used for deploying to Azure Kubernetes Service (AKS), Google Kubernetes Engine (GKE), Amazon Elastic Kubernetes Service (EKS), or clusters from any other cloud providers.

To set up multicloud deployment, create an environment and then add your Kubernetes resources associated with namespaces of Kubernetes clusters.

The generic provider approach based on existing service account works with clusters from any cloud provider, including Azure. The benefit of using the Azure Kubernetes Service option instead is that it creates new ServiceAccount and RoleBinding objects (instead of reusing an existing ServiceAccount) so that the newly created RoleBinding object can limit the operations of the ServiceAccount to the chosen namespace only.

When you use the generic provider approach, make sure that a RoleBinding exists, which grants permissions in the edit ClusterRole to the desired service account. You need to grant permissions to the right services account so that the service account can be used by Azure Pipelines to create objects in the chosen namespace.

Parallel deployments to multiple clouds

The following example shows how to do parallel deployments to clusters in multiple clouds. In this example, there are deployments to the AKS, GKE, EKS, and OpenShift clusters. These four namespaces are associated with Kubernetes resources under the contoso environment.

trigger:
- main

jobs:
- deployment:
  displayName: Deploy to AKS
  pool:
    vmImage: ubuntu-latest
  environment: contoso.aksnamespace
  strategy:
    runOnce:
      deploy:
        steps:
        - checkout: self
        - task: KubernetesManifest@0
          displayName: Deploy to Kubernetes cluster
          inputs:
            action: deploy
            kubernetesServiceConnection: serviceConnection #replace with your service connection
            namespace: aksnamespace
            manifests: manifests/*
- deployment:
  displayName: Deploy to GKE
  pool:
    vmImage: ubuntu-latest
  environment: contoso.gkenamespace
  strategy:
    runOnce:
      deploy:
        steps:
        - checkout: self
        - task: KubernetesManifest@0
          displayName: Deploy to Kubernetes cluster
          inputs:
            action: deploy
            kubernetesServiceConnection: serviceConnection #replace with your service connection
            namespace: gkenamespace
            manifests: manifests/*
- deployment:
  displayName: Deploy to EKS
  pool:
    vmImage: ubuntu-latest
  environment: contoso.eksnamespace
  strategy:
    runOnce:
      deploy:
        steps:
        - checkout: self
        - task: KubernetesManifest@0
          displayName: Deploy to Kubernetes cluster
          inputs:
            action: deploy
            kubernetesServiceConnection: serviceConnection #replace with your service connection
            namespace: eksnamespace
            manifests: manifests/*
- deployment:
  displayName: Deploy to OpenShift
  pool:
    vmImage: ubuntu-latest
  environment: contoso.openshiftnamespace
  strategy:
    runOnce:
      deploy:
        steps:
        - checkout: self
        - task: KubernetesManifest@0
          displayName: Deploy to Kubernetes cluster
          inputs:
            action: deploy
            kubernetesServiceConnection: serviceConnection #replace with your service connection
            namespace: openshiftnamespace
            manifests: manifests/*
- deployment:
  displayName: Deploy to DigitalOcean
  pool:
    vmImage: ubuntu-latest
  environment: contoso.digitaloceannamespace
  strategy:
    runOnce:
      deploy:
        steps:
        - checkout: self
        - task: KubernetesManifest@0
          displayName: Deploy to Kubernetes cluster
          inputs:
            action: deploy
            kubernetesServiceConnection: serviceConnection #replace with your service connection
            namespace: digitaloceannamespace
            manifests: manifests/*