master | dev | |
---|---|---|
Coverage | ||
Unit Test | ||
.NETCore Windows | ||
.NETCore Linux | ||
Java Windows | ||
Java Linux | ||
Node JS Windows | ||
Node JS Linux | ||
PowerShell Windows | ||
Python Linux |
The Azure Functions action is used in a GitHub Actions workflow to deploy packaged project code to an existing function app hosted in Azure. Using this action, you can create continuous workflow automation that builds, authenticates, and deploys code to your function app when you make changes in the GitHub repository.
The Azure Functions action is defined in this action.yml file. To learn about specific parameters, see the Reference section in this readme.
Important
If you're looking for a GitHub Action to deploy your customized container image to Azure Functions, instead use functions-container-action.
You can create a workflow for your function app deployments in one of these ways:
-
Workflow templates: choose a workflow template that matches your function app's code language, operating system, and hosting plan. Add this workflow (.yaml) file to the
.github/workflows/
folder of the repository that contains your project code. -
Azure portal: you can create a workflow (.yaml) file directly in your repository from your app's Deployment center page in the portal by following the detailed instructions in Continuous delivery by using GitHub Actions.
Keep these considerations in mind when working with the azure-functions
action in your workflows:
-
These sample templates require updates when deploying apps that run in a Flex Consumption plan. For more information, see GitHub Action parameters.
-
When using run-from-package deployment, apps can use external storage on a blob container, which is specified using the app setting
WEBSITE_RUN_FROM_PACKAGE=<URL>
. In this case, the<URL>
points to the external blob storage container. This is the default deployment for apps running on Linux in a Consumption plan. To protect the deployment package, the container should require private access. Managed identities are recommended, but shared access signature (SAS) tokens can also be used. There are specific app settings required when using managed identities to access the deployment container. For more information, see Fetch a package from Azure Blob Storage using a managed identity. -
If you have a non-.NET project that includes an explicit binding reference (extensions.csproj), the Functions Action ignores this file and doesn't build the extensions project. For more information, see Binding extensions projects.
-
The remove additional files at destination functionality provided by a
scm
(Kudu) deployment isn't supported by the deployment method used in this action. When you use zip deploy or one deploy for Flex Consumption plans, all files from a previous deployment are removed or are updated with the latest files from the deployment package. Any other files and directories outside of the deployment, such as added using FTP or created by your app during runtime, are preserved. If you're usingscm
with a package deployment, you must handle the removal of previously deployed files outside of the Functions action.
You must choose how the action authenticates with Azure when deploying code from GitHub to your function app. There are three supported authentication methods:
Method | Description |
---|---|
OpenID Connect (OIDC) | Recommended. This is the most secure authentication option. Supports role-based access control (RBAC) for accessing Azure resources. Leverages user-assigned managed identities, which means that secrets are managed by Azure. |
Azure service principal | Not recommended. GitHub uses service principal credentials to access Azure resources using RBAC. You must manage the service principal secrets in GitHub. |
Publish profile | Not recommended. Credentials (username and password) are stored in GitHub and used to access the scm endpoint using HTTP basic authentication during deployment. This will require you to enable basic authentication for your app. |
These are special considerations for certain hosting scenarios:
- When your app is hosted in a sovereign cloud or on Azure Stack Hub, make sure to include the relevant
environment
andaudience
parameters for theazure/login
action when using OIDC or service principal authentication. - Publish profile as an authentication method is not supported when your app is hosted on Linux in a Consumption plan and the project contains an executable file, such as a custom handler, or
chrome
in Puppeteer/Playwright. - OIDC authentication uses Workload identity federation in Azure, which only supports user-assigned managed identities.
When using Open ID Connect (OIDC) for authentication, you configure a user-assigned managed identity in Azure to allow your GitHub to use this identity from the context of your workflow. The workflow can then authenticate with Azure without the need for credentials. You must also grant the managed identity permissions to deploy to your function app, which is done using a role assignment.
Tip
This method is the most secure and recommended for those with permissions to configure identity.
To configure your workflow to use OIDC authentication:
-
Copy down the values for Client ID, Subscription ID, and Directory (tenant) ID for the new user-assigned managed identity resource.
-
Assign the role [
Website Contributor'](https://learn.microsoft.com/azure/role-based-access-control/built-in-roles/web-and-mobile#website-contributor)
to your user-assigned managed identity, scoped to the function app you want to deploy to. -
Create a federated credential with the following settings:
- Federated credential scenario:
Configure a GitHub issued token to impersonate this application and deploy to Azure
- Organization:
<GITHUB_ORGANIZATION>
- Repository:
<REPO_CONTAINING_FUNCTIONS_PROJECT>
- Entity type:
Branch
- Branch:
<BRANCH_NAME>
Replace these placeholders with the information for your GitHub repository.
- Federated credential scenario:
Tip
The Entity
in the OIDC definition is what triggers the workflow to fetch the OIDC token. Our samples assume you want to trigger the workflow on pushes to main
. To use a different trigger, see our entity type examples.
-
Create these variables in your repository using the values you recorded in step 2:
- AZURE_CLIENT_ID:
<CLIENT_ID>
- AZURE_TENANT_ID:
<DIRECTORY_TENANT_ID>
- AZURE_SUBSCRIPTION_ID:
<SUBSCRIPTION_ID>
- AZURE_CLIENT_ID:
-
In the job containing the
checkout
action (in our samples, thebuild
job), add the following permissions:id-token: write
contents: read
Note
id-token: write
is required for GitHub's OIDC provider to fetch an OIDC token. contents: read
is required to checkout a private repo. Learn more about OIDC permissions.
-
Add an
azure/login
action in the step before your Azure Functions action. The new action must have these parameters, which map to the new variables you just added to your repository:client-id
: ${{ vars.AZURE_CLIENT_ID }}tenant-id
: ${{ vars.AZURE_TENANT_ID }}subscription-id
: ${{ vars.AZURE_SUBSCRIPTION_ID }}
-
In the job containing the Azure Functions action (in our samples, the
deploy
job) add theid-token: write
permission.
When using OIDC authentication, the jobs
section of your workflow looks something like this:
# Deploy to an app on the Flex Consumption plan using OIDC authentication
jobs:
build:
runs-on: ubuntu-latest
permissions:
id-token: write # Required for OIDC
contents: read # Required for actions/checkout
steps:
# ...checkout your repository
# ...required build steps for your language
# ...upload your build artifact
deploy:
runs-on: ubuntu-latest
needs: build
permissions:
id-token: write # Required for OIDC
steps:
# ...download your build artifact
- name: 'Log in to Azure with AZ CLI'
uses: azure/login@v2
with:
client-id: ${{ vars.AZURE_CLIENT_ID }} # Required to log in with OIDC
tenant-id: ${{ vars.AZURE_TENANT_ID }} # Required to log in with OIDC
subscription-id: ${{ vars.AZURE_SUBSCRIPTION_ID }} # Required to log in with OIDC
- name: 'Run the Azure Functions action'
uses: Azure/functions-action@v1
id: deploy-to-function-app
with:
app-name: ${{ env.AZURE_FUNCTIONAPP_NAME }}
package: '${{ env.AZURE_FUNCTIONAPP_PROJECT_PATH }}'
You can alternatively use a service principal, which requires you to manage secrets. You must configure the workflow with these secrets, and then it can use them to authenticate with Azure.
Important
When possible, you should use OIDC for authentication instead of service principal-based authentication.
To configure your workflow to use a service principal for authentication:
-
If you don't already have it installed, download Azure CLI and run
az login
to sign in with your Azure credentials. -
Run this Azure CLI command:
az ad sp create-for-rbac --name "myApp" --role "Website Contributor" \ --scopes /subscriptions/<SUBSCRIPTION_ID>/resourceGroups/<RESOURCE_GROUP>/providers/Microsoft.Web/sites/<APP_NAME> \ --json-auth
In this example, Replace
<SUBSCRIPTION_ID>
,<RESOURCE_GROUP>
, and<APP_NAME>
with the names of your subscription, resource group, and Azure function app. The command should return JSON output like this:{ "clientId": "<GUID>", "clientSecret": "<GUID>", "subscriptionId": "<GUID>", "tenantId": "<GUID>", (...) }
-
Copy this JSON response output, which is the credential you provide to GitHub for authentication.
Warning
Keep this credential safe. It provides Website Contributor
role access to your function app.
-
In your GitHub Repository, select Settings > Secrets > Add a new secret, name the secret something like
AZURE_RBAC_CREDENTIALS
, and paste in JSON credentials of the service principal. -
Add the
azure/login
action as a step prior to the Azure Functions action:- Make sure to include the parameter
cred-id
, which maps to your recently created repository secretAZURE_RBAC_CREDENTIALS
. - Make sure you don't also have any publish profiles in your workflow, which would be in the
publish-profile
parameter of the Azure Functions action.
- Make sure to include the parameter
When you use a service principal with RBAC, the jobs
section of your workflow looks something like this:
# Deploy to an app on the Flex Consumption plan using a service principal with RBAC as authentication
jobs:
build:
runs-on: ubuntu-latest
steps:
# ...checkout your repository
# ...required build steps for your language
# ...upload your build artifact
deploy:
runs-on: ubuntu-latest
needs: build
steps:
# ...download your build artifact
- name: 'Log in to Azure with AZ CLI'
uses: azure/login@v2
with:
cred-id: ${{ secrets.AZURE_RBAC_CREDENTIALS }}
- name: 'Run the Azure Functions action'
uses: Azure/functions-action@v1
id: deploy-to-function-app
with:
app-name: ${{ env.AZURE_FUNCTIONAPP_NAME }}
package: '${{ env.AZURE_FUNCTIONAPP_PROJECT_PATH }}'
A publish profile contains plain-text secrets that authenticate with your function app using basic authentication with the scm
HTTP endpoint.
Warning
Publish profile authentication uses a shared secret which you must manage. It also requires you to enable publishing credential access to the app, which is off by default and is not recommended. You should instead use a more secure option like OIDC authentication.
To configure your workflow using the publish profile:
-
In the Azure portal, locate your function app.
-
Make sure that Basic authentication is enabled in the
scm
endpoint in your app under Settings > Configuration > SCM Basic Auth Publishing Credentials. -
In the Overview blade, select Get publish profile and download the .PublishSettings file, which contains the plain-text publishing credentials for your
scm
endpoint. -
Open the .PublishSettings file and copy the XML file contents. Delete or secure this secrets file when you're done.
-
In your GitHub Repository, select Settings > Secrets > Add a new secret, name the secret AZURE_FUNCTIONAPP_PUBLISH_PROFILE, and paste in the XML profile file contents.
-
Make sure that your workflow isn't using the
azure/login
action. -
Include the
publish-profile
parameter in the Azure Functions action, referencing the AZURE_FUNCTIONAPP_PUBLISH_PROFILE secret.When using a publish profile, the
jobs
section of your workflow looks something like this:# Deploy to an app on the Flex Consumption plan using RBAC with a service principal as authentication jobs: build: runs-on: ubuntu-latest steps: # ...checkout your repository # ...required build steps for your language # ...upload your build artifact deploy: runs-on: ubuntu-latest needs: build steps: # ...download your build artifact - name: 'Run the Azure Functions action' uses: Azure/functions-action@v1 id: deploy-to-function-app with: app-name: ${{ env.AZURE_FUNCTIONAPP_NAME }} package: '${{ env.AZURE_FUNCTIONAPP_PROJECT_PATH }}' sku: `flexconsumption` # Parameter required when using a publish profile with Flex Consumption publish-profile: ${{ secrets.AZURE_FUNCTIONAPP_PUBLISH_PROFILE }}
This section describes the input parameters supported by the Azure Functions action. The parameters you use depends how your function app is hosted on Azure. For example, the Flex Consumption plan uses different parameters than the other plans to enable a remote build. If you intend to deploy to the Container Apps plan, use functions-container-action
instead.
Parameters required on all hosting plans:
Parameter | Description |
---|---|
app-name | The function app name on Azure. For example, when your app's site URL is https://your-site-name.azurewebsites.net/ , then app-name is your-site-name . |
package | This is the path to your project in the repo being published. By default, this value is set to . , which means all files and folders in the GitHub repository are deployed. |
Parameters used with the Flex Consumption plan:
Parameter | Description |
---|---|
sku | Required only when authenticating a Flex Consumption plan app deployment using publish-profile . In this case, you must set sku to flexconsumption . When publishing to a Flex Consumption plan app using OIDC or service principle authentication with RBAC credentials, the action can resolve the value. Other plans also don't require this parameter. |
remote-build | When set to true , enables a build action in the scm (Kudu) site. By default, this parameter is set to false . For a Flex Consumption plan app, an Oryx build is always performed during a remote build in Flex Consumption; don't set scm-do-build-during-deployment or enable-oryx-build. |
Parameters used with the Consumption, Elastic Premium, and App Service (Dedicated) plans:
Parameter | Description |
---|---|
scm-do-build-during-deployment | Requests the scm (Kudu) site to perform predeployment operations, such as npm install or dotnet build /dotnet publish . This is equivalent to using the SCM_DO_BUILD_DURING_DEPLOYMENT app setting. A remote build isn't required for package-based deployments, so this is set to false by default. If for some reason you don't want to use a deployment package and instead need to use Kudu or KuduLite, set this to true to have Kudu build your project during deployment. For more information, see remote build. |
enable-oryx-build | Requests that a remote scm (Kudu) build be done using the Oryx build system. By default, this is set to false . To have Oryx to resolve your dependencies and build your project instead of the workflow, set both scm-do-build-during-deployment and enable-oryx-build to true . Only supported for Linux function apps. |
Optional parameters for all hosting plans:
Parameter | Description |
---|---|
slot-name | Specifies a named slot as the deployment target. By default, this value isn't set, which means the action deploys to your production slot. When this setting resolves to a named slot, make sure that publish-profile also contains the credentials for the target slot instead of the production slot. Currently not supported in a Flex Consumption plan. |
publish-profile | The plain-text credentials used to access the scm endpoint using HTTP basic authentication during deployment. This must contain the XML contents of your .PublishSettings file. To use this authentication method, see Publish profile (HTTP basic) authentication. We highly recommend setting the content in GitHub secrets since it contains sensitive information such as your site URL, username, and password. When the publish profile is rotated in your function app, you also need to update the GitHub secret. Otherwise, a 401 error occurs when accessing the /api/settings endpoint. |
respect-pom-xml | Allow the GitHub Action to derive the Java function app's artifact from pom.xml for deployments. By default, it's set to false , which means the package parameter needs to point to your Java function app's artifact, such as ./target/azure-functions/<FUNCTION_APP_NAME> . It's recommended to set package to . and respect-pom-xml to true when deploying Java function apps. |
respect-funcignore | Allow the GitHub Action to parse your .funcignore file and exclude files and folders defined in it. By default, this value is set to false . If your GitHub repo contains .funcignore file and you want to exclude certain paths (for example, text editor configs .vscode/, Python virtual environment .venv/), we recommend setting this to true . |
By default, non-.NET projects use extension bundles to include the .NET NuGet packages potentially required by binding extensions in your app.
In some cases, such as when using a prerelease or specific version of an extension, you're required to explicitly install extensions, which is done in an extensions.csproj file in the root folder. The Functions Action doesn't build the extension.csproj file.
If you can't convert to using extension bundles, you must instead include these steps before the function-action
step in your workflow to build the extensions project:
# Adding DOTNET_VERSION to your workflow's environment variables is recommeded
...
- name: 'Set up .NET version: ${{ env.DOTNET_VERSION }}'
uses: actions/setup-dotnet@v4
with:
dotnet-version: ${{ env.DOTNET_VERSION }}
- name: 'Build the extensions project'
run: dotnet build --output ./bin
The Azure Functions action itself has no dependencies. However, you might need the following actions when constructing an end-to-end GitHub workflow similar to our samples.
- Checkout: Check out your Git repository content into GitHub Actions agent.
- Azure Login: Sign in with your Azure credentials when authenticating with OIDC or service principal.
- Environment setup actions:
- Setup DotNet: Build your DotNet core function app or function app extensions.
- Setup Node: Resolve Node function app dependencies using npm.
- Setup Python: Resolve Python function app dependencies using pip.
- Setup Java: Resolve Java function app dependencies using maven.
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 (for example, status check, comment). Follow the instructions provided by the bot. You'll 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 [email protected] with any other questions or comments.