This action is used to deploy manifests to Kubernetes clusters. It requires that the cluster context be set earlier in the workflow by using either the Azure/aks-set-context action or the Azure/k8s-set-context action. It also requires Kubectl to be installed (you can use the Azure/setup-kubectl action).
If you are looking to automate your workflows to deploy to Azure Web Apps and Azure Web App for Containers, consider using Azure/webapps-deploy action.
This action requires the following permissions from your workflow:
permissions:
id-token: write
contents: read
actions: readFollowing are the key capabilities of this action:
-
Artifact substitution: Takes a list of container images which can be specified along with their tags or digests. They are substituted into the non-templatized version of manifest files before applying to the cluster to ensure that the right version of the image is pulled by the cluster nodes.
-
Object stability checks: Rollout status is checked for the Kubernetes objects deployed. This is done to incorporate stability checks while computing the action status as success/failure.
-
Secret handling: The secret names specified as inputs in the action are used to augment the input manifest files with imagePullSecrets values before deploying to the cluster. Also, checkout the Azure/k8s-create-secret action for creation of generic or docker-registry secrets in the cluster.
-
Deployment strategy Supports both canary and blue-green deployment strategies
-
Canary strategy: Workloads suffixed with '-baseline' and '-canary' are created. There are two methods of traffic splitting supported:
- Service Mesh Interface: Service Mesh Interface abstraction allows for plug-and-play configuration with service mesh providers such as Linkerd and Istio. Meanwhile, this action takes away the hard work of mapping SMI's TrafficSplit objects to the stable, baseline and canary services during the lifecycle of the deployment strategy. Service mesh based canary deployments using this action are more accurate as service mesh providers enable granular percentage traffic split (via service registry and sidecar containers injected into pods alongside application containers).
- Only Kubernetes (no service mesh): In the absence of service mesh, while it may not be possible to achieve exact percentage split at the request level, it is still possible to perform canary deployments by deploying -baseline and -canary workload variants next to the stable variant. The service routes requests to pods of all three workload variants as the selector-label constraints are met (KubernetesManifest will honor these when creating -baseline and -canary variants). This achieves the intended effect of routing only a portion of total requests to the canary.
-
Blue-Green strategy: Choosing blue-green strategy with this action leads to creation of workloads suffixed with '-green'. An identified service is one that is supplied as part of the input manifest(s) and targets a workload in the supplied manifest(s). There are three route-methods supported in the action:
- Service route-method: Identified services are configured to target the green deployments.
- Ingress route-method: Along with deployments, new services are created with '-green' suffix (for identified services), and the ingresses are in turn updated to target the new services.
- SMI route-method: A new TrafficSplit object is created for each identified service. The TrafficSplit object is updated to target the new deployments. This works only if SMI is set up in the cluster.
Traffic is routed to the new workloads only after the time provided as
version-switch-bufferinput has passed. Thepromoteaction creates workloads and services with new configurations but without any suffix.rejectroutes traffic back to the old workloads and deletes the '-green' workloads.
-
| Action inputs | Description |
|---|---|
| action (Required) |
Acceptable values: deploy/promote/reject. Promote or reject actions are used to promote or reject canary/blue-green deployments. Sample YAML snippets are provided below for guidance. |
| manifests (Required) |
Path to the manifest files to be used for deployment. These can also be directories containing manifest files, in which case, all manifest files in the referenced directory at every depth will be deployed, or URLs to manifest files (like https://raw.githubusercontent.com/kubernetes/website/main/content/en/examples/controllers/nginx-deployment.yaml). Files and URLs not ending in .yml or .yaml will be ignored. |
| strategy (Required) |
Acceptable values: basic/canary/blue-green. Default value: basic Deployment strategy to be used while applying manifest files on the cluster. basic - Template is force applied to all pods when deploying to cluster. NOTE: Can only be used with action == deploy canary - Canary deployment strategy is used when deploying to the cluster. blue-green - Blue-Green deployment strategy is used when deploying to cluster. |
| namespace (Optional) | Namespace within the cluster to deploy to. |
| images (Optional) |
Fully qualified resource URL of the image(s) to be used for substitutions on the manifest files. This multiline input accepts specifying multiple artifact substitutions in newline separated form. For example:In this example, all references to contosodemo.azurecr.io/foo and contosodemo.azurecr.io/bar are searched for in the image field of the input manifest files. For the matches found, the tags test1 and test2 are substituted. |
| imagepullsecrets (Optional) |
Multiline input where each line contains the name of a docker-registry secret that has already been setup within the cluster. Each of these secret names are added under imagePullSecrets field for the workloads found in the input manifest files |
| pull-images (Optional) |
Acceptable values: true/false Default value: true Switch whether to pull the images from the registry before deployment to find out Dockerfile's path in order to add it to the annotations |
| traffic-split-method (Optional) |
Acceptable values: pod/smi. Default value: pod SMI: Percentage traffic split is done at request level using service mesh. Service mesh has to be setup by cluster admin. Orchestration of TrafficSplit objects of SMI is handled by this action. Pod: Percentage split not possible at request level in the absence of service mesh. Percentage input is used to calculate the replicas for baseline and canary as a percentage of replicas specified in the input manifests for the stable variant. |
| traffic-split-annotations (Optional) |
Annotations in the form of key/value pair to be added to TrafficSplit. |
| percentage (Optional but required if strategy is canary) |
Used to compute the number of replicas of '-baseline' and '-canary' variants of the workloads found in manifest files. For the specified percentage input, if (percentage * numberOfDesirerdReplicas)/100 is not a round number, the floor of this number is used while creating '-baseline' and '-canary'. For example, if Deployment hello-world was found in the input manifest file with 'replicas: 4' and if 'strategy: canary' and 'percentage: 25' are given as inputs to the action, then the Deployments hello-world-baseline and hello-world-canary are created with 1 replica each. The '-baseline' variant is created with the same image and tag as the stable version (4 replica variant prior to deployment) while the '-canary' variant is created with the image and tag corresponding to the new changes being deployed |
| baseline-and-canary-replicas (Optional and relevant only if strategy is canary and traffic-split-method is smi) |
The number of baseline and canary replicas. Percentage traffic split is controlled in the service mesh plane, the actual number of replicas for canary and baseline variants could be controlled independently of the traffic split. For example, assume that the input Deployment manifest desired 30 replicas to be used for stable and that the following inputs were specified for the action     strategy: canaryIn this case, stable variant will receive 80% traffic while baseline and canary variants will receive 10% each (20% split equally between baseline and canary). However, instead of creating baseline and canary with 3 replicas each, the explicit count of baseline and canary replicas is honored. That is, only 1 replica each is created for baseline and canary variants. |
| route-method (Optional and relevant only if strategy is blue-green) |
Acceptable values: service/ingress/smi. Default value: service. Traffic is routed based on this input. Service: Service selector labels are updated to target '-green' workloads. Ingress: Ingress backends are updated to target the new '-green' services which in turn target '-green' deployments. SMI: A TrafficSplit object is created for each required service to route traffic to new workloads. |
| version-switch-buffer (Optional and relevant only if strategy is blue-green) |
Acceptable values: 1-300. Default value: 0. Waits for the given input in minutes before routing traffic to '-green' workloads. |
| private-cluster (Optional and relevant only using K8's deploy for a cluster with private cluster enabled) |
Acceptable values: true, false Default value: false. |
| force (Optional) |
Deploy when a previous deployment already exists. If true then '--force' argument is added to the apply command. Using '--force' argument is not recommended in production. |
| annotate-resources (Optional) |
Acceptable values: true/false Default value: true Switch whether to annotate the resources or not. If set to false all annotations are skipped completely. |
| annotate-namespace (Optional) |
Acceptable values: true/false Default value: true Switch whether to annotate the namespace resources object or not. Ignored when annotate-resources is set to false. |
| skip-tls-verify (Optional) |
Acceptable values: true/false Default value: false True if the insecure-skip-tls-verify option should be used |
| resource-type (Optional) | Acceptable values: `Microsoft.ContainerService/managedClusters` (default), 'Microsoft.ContainerService/fleets' |
- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
manifests: |
dir/manifestsDirectory
images: 'contoso.azurecr.io/myapp:${{ event.run_id }}'
imagepullsecrets: |
image-pull-secret1
image-pull-secret2- uses: Azure/k8s-deploy@v5
with:
resource-group: yourResourceGroup
name: yourClusterName
action: deploy
strategy: basic
private-cluster: true
manifests: |
manifests/azure-vote-backend-deployment.yaml
manifests/azure-vote-backend-service.yaml
manifests/azure-vote-frontend-deployment.yaml
manifests/azure-vote-frontend-service.yaml
images: |
registry.azurecr.io/containername- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
images: 'contoso.azurecr.io/myapp:${{ event.run_id }}'
imagepullsecrets: |
image-pull-secret1
image-pull-secret2
manifests: |
deployment.yaml
service.yaml
dir/manifestsDirectory
strategy: canary
action: deploy
percentage: 20To promote/reject the canary created by the above snippet, the following YAML snippet could be used:
- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
images: 'contoso.azurecr.io/myapp:${{ event.run_id }}'
imagepullsecrets: |
image-pull-secret1
image-pull-secret2
manifests: |
deployment.yaml
service.yaml
dir/manifestsDirectory
strategy: canary
action: promote # substitute reject if you want to reject- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
images: 'contoso.azurecr.io/myapp:${{ event.run_id }}'
imagepullsecrets: |
image-pull-secret1
image-pull-secret2
manifests: |
deployment.yaml
service.yaml
dir/manifestsDirectory
strategy: canary
action: deploy
traffic-split-method: smi
percentage: 20
baseline-and-canary-replicas: 1To promote/reject the canary created by the above snippet, the following YAML snippet could be used:
- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
images: 'contoso.azurecr.io/myapp:${{ event.run_id }} '
imagepullsecrets: |
image-pull-secret1
image-pull-secret2
manifests: |
deployment.yaml
service.yaml
dir/manifestsDirectory
strategy: canary
traffic-split-method: smi
action: reject # substitute promote if you want to promote- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
images: 'contoso.azurecr.io/myapp:${{ event.run_id }}'
imagepullsecrets: |
image-pull-secret1
image-pull-secret2
manifests: |
deployment.yaml
service.yaml
ingress.yml
strategy: blue-green
action: deploy
route-method: ingress # substitute with service/smi as per need
version-switch-buffer: 15To promote/reject the green workload created by the above snippet, the following YAML snippet could be used:
- uses: Azure/k8s-deploy@v5
with:
namespace: 'myapp'
images: 'contoso.azurecr.io/myapp:${{ event.run_id }}'
imagepullsecrets: |
image-pull-secret1
image-pull-secret2
manifests: |
deployment.yaml
service.yaml
ingress.yml
strategy: blue-green
route-method: ingress # should be the same as the value when action was deploy
action: promote # substitute reject if you want to rejectFollowing are a few examples of not just this action, but how this action could be used along with other container and k8s related actions for building images and deploying objects onto k8s clusters:
on: [push]
jobs:
build:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: Azure/docker-login@v1
with:
login-server: contoso.azurecr.io
username: ${{ secrets.REGISTRY_USERNAME }}
password: ${{ secrets.REGISTRY_PASSWORD }}
- run: |
docker build . -t contoso.azurecr.io/k8sdemo:${{ github.sha }}
docker push contoso.azurecr.io/k8sdemo:${{ github.sha }}
- uses: azure/setup-kubectl@v4
# Set the target AKS cluster.
- uses: Azure/aks-set-context@v4
with:
creds: '${{ secrets.AZURE_CREDENTIALS }}'
cluster-name: contoso
resource-group: contoso-rg
- uses: Azure/k8s-create-secret@v4
with:
container-registry-url: contoso.azurecr.io
container-registry-username: ${{ secrets.REGISTRY_USERNAME }}
container-registry-password: ${{ secrets.REGISTRY_PASSWORD }}
secret-name: demo-k8s-secret
- uses: Azure/k8s-deploy@v5
with:
action: deploy
manifests: |
manifests/deployment.yml
manifests/service.yml
images: |
demo.azurecr.io/k8sdemo:${{ github.sha }}
imagepullsecrets: |
demo-k8s-secreton: [push]
jobs:
build:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: Azure/docker-login@v1
with:
login-server: contoso.azurecr.io
username: ${{ secrets.REGISTRY_USERNAME }}
password: ${{ secrets.REGISTRY_PASSWORD }}
- run: |
docker build . -t contoso.azurecr.io/k8sdemo:${{ github.sha }}
docker push contoso.azurecr.io/k8sdemo:${{ github.sha }}
- uses: azure/setup-kubectl@v4
- uses: Azure/k8s-set-context@v4
with:
kubeconfig: ${{ secrets.KUBE_CONFIG }}
- uses: Azure/k8s-create-secret@v4
with:
container-registry-url: contoso.azurecr.io
container-registry-username: ${{ secrets.REGISTRY_USERNAME }}
container-registry-password: ${{ secrets.REGISTRY_PASSWORD }}
secret-name: demo-k8s-secret
- uses: Azure/k8s-deploy@v4
with:
action: deploy
manifests: |
manifests/deployment.yml
manifests/service.yml
images: |
demo.azurecr.io/k8sdemo:${{ github.sha }}
imagepullsecrets: |
demo-k8s-secretWe can use this image in other workflows once built.
on: [push]
env:
NAMESPACE: demo-ns2
jobs:
build:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: Azure/docker-login@v1
with:
login-server: contoso.azurecr.io
username: ${{ secrets.REGISTRY_USERNAME }}
password: ${{ secrets.REGISTRY_PASSWORD }}
- run: |
docker build . -t contoso.azurecr.io/k8sdemo:${{ github.sha }} --label dockerfile-path=https://github.com/${{github.repo}}/blob/${{github.sha}}/Dockerfile
docker push contoso.azurecr.io/k8sdemo:${{ github.sha }}on: [push]
env:
NAMESPACE: demo-ns2
jobs:
deploy:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@master
- uses: Azure/docker-login@v1
with:
login-server: contoso.azurecr.io
username: ${{ secrets.REGISTRY_USERNAME }}
password: ${{ secrets.REGISTRY_PASSWORD }}
- uses: azure/setup-kubectl@v4
# Set the target AKS cluster.
- uses: Azure/aks-set-context@v4
with:
creds: '${{ secrets.AZURE_CREDENTIALS }}'
cluster-name: contoso
resource-group: contoso-rg
- uses: Azure/k8s-create-secret@v4
with:
namespace: ${{ env.NAMESPACE }}
container-registry-url: contoso.azurecr.io
container-registry-username: ${{ secrets.REGISTRY_USERNAME }}
container-registry-password: ${{ secrets.REGISTRY_PASSWORD }}
secret-name: demo-k8s-secret
- uses: azure/k8s-bake@v3
with:
renderEngine: 'helm'
helmChart: './aks-helloworld/'
overrideFiles: './aks-helloworld/values-override.yaml'
overrides: |
replicas:2
helm-version: 'latest'
id: bake
- uses: Azure/k8s-deploy@v5
with:
action: deploy
manifests: ${{ steps.bake.outputs.manifestsBundle }}
images: |
contoso.azurecr.io/k8sdemo:${{ github.sha }}
imagepullsecrets: |
demo-k8s-secret- Environment variable
HELM_CHART_PATHSis a list of helmchart files expected by k8s-deploy - it will be populated automatically if you are using k8s-bake to generate the manifests. - Use script to build image and add dockerfile-path label to it. The value expected is the link to the dockerfile: https://github.com/${{github.repo}}/blob/${{github.sha}}/Dockerfile. If your dockerfile is in the same repo and branch where the workflow is run, it can be a relative path and it will be converted to a link for traceability.
- Run docker login action for each image registry - in case image build and image deploy are two distinct jobs in the same or separate workflows.
This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.
k8s-deploy is an open source project that is not covered by the Microsoft Azure support policy. Please search open issues here, and if your issue isn't already represented please open a new one. The project maintainers will respond to the best of their abilities.