Deploy a Function from a Network-Restricted Git Repository ​
NOTE
Use this approach when your Git repository is hosted on a server that is not reachable from the Kyma cluster network, for example, a GitHub Enterprise instance behind a corporate firewall. If your cluster can reach the repository, use Git-sourced Functions instead.
This tutorial shows how to use a GitHub Actions workflow to deploy a Kyma Function whenever you push changes to your repository. Instead of letting the cluster pull source code from Git, the workflow pushes the Function code directly to the cluster as an inline Function. The cluster does not need to reach the Git repository.
Prerequisites ​
- You have the Serverless module added.
- You have a GitHub repository with GitHub Actions enabled.
- You have
kubectlinstalled locally. - You have cluster admin access to create service accounts and roles.
How It Works ​
On every push to the configured branch, GitHub Actions:
- Checks out your repository code.
- Installs Kyma CLI using
kyma-project/setup-kyma-cli. - Authenticates to the Kyma cluster using a kubeconfig stored as a GitHub Secret.
- Deploys the Function as an inline Function using
kyma function create nodejs.
NOTE
The kyma function commands are Kyma CLI extensions served by the Serverless module installed on your cluster. The GitHub Actions runner must be able to reach the Kyma API server to discover and run these commands.
Procedure ​
Step 1: Create a Service Account and Kubeconfig ​
Create a dedicated Kubernetes service account with the minimum permissions required to manage Functions in your namespace.
export NAMESPACE={YOUR_NAMESPACE}
kubectl -n $NAMESPACE create serviceaccount function-deployer
kubectl -n $NAMESPACE create role function-deployer-role \
--verb=get,list,create,update,patch,delete \
--resource=functions.serverless.kyma-project.io
kubectl -n $NAMESPACE create rolebinding function-deployer-binding \
--role=function-deployer-role \
--serviceaccount=$NAMESPACE:function-deployerGenerate a token and save it as a kubeconfig file:
TOKEN=$(kubectl -n $NAMESPACE create token function-deployer --duration=8760h)
CLUSTER=$(kubectl config view --minify -o jsonpath='{.clusters[0].name}')
kubectl config set-credentials function-deployer --token=$TOKEN
kubectl config set-context function-deployer-ctx \
--cluster=$CLUSTER \
--user=function-deployer \
--namespace=$NAMESPACE
CURRENT_CONTEXT=$(kubectl config current-context)
kubectl config use-context function-deployer-ctx
kubectl config view --minify --flatten > function-deployer-kubeconfig.yaml
# Restore your original context
kubectl config use-context $CURRENT_CONTEXTWARNING
The generated token may expire earlier than requested. Managed Kyma clusters cap the token lifetime (often 24 to 48 hours). If the workflow starts failing with authentication errors, the token has expired. Re-run Step 1 to generate a new one and update the GitHub Secret. Store the kubeconfig securely and rotate it regularly.
Step 2: Store the Kubeconfig as a GitHub Secret ​
- Open your GitHub repository and go to Settings > Secrets and variables > Actions.
- Select New repository secret.
- Set the name to
KYMA_KUBECONFIGand paste the contents offunction-deployer-kubeconfig.yamlas the value. - Select Add secret.
Step 3: Add the GitHub Actions Workflow ​
Create the file .github/workflows/deploy-function.yml in your repository:
name: Deploy Kyma Function
on:
push:
branches: [main]
paths:
- 'function/**'
jobs:
deploy:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: kyma-project/setup-kyma-cli@v1
with:
version: stable
- name: Deploy Function
env:
KUBECONFIG_CONTENT: ${{ secrets.KYMA_KUBECONFIG }}
run: |
echo "$KUBECONFIG_CONTENT" > /tmp/kubeconfig.yaml
kyma --kubeconfig /tmp/kubeconfig.yaml function create nodejs my-function \
--namespace {YOUR_NAMESPACE} \
--source function/handler.js \
--dependencies function/package.jsonNOTE
Replace {YOUR_NAMESPACE} and my-function with your actual namespace and Function name. Adjust paths, branches, --source, and --dependencies to match your repository structure. To use Python, replace nodejs with python. For a specific Node.js version, add --version 22 or --version 26.
Step 4: Verify the Deployment ​
After pushing to the main branch, check that the Function reached the RUNNING state:
kubectl get functions my-function -n {YOUR_NAMESPACE}Expected output:
NAME CONFIGURED BUILT RUNNING RUNTIME VERSION AGE
my-function True True True nodejs24 1 60sSecurity Considerations ​
- Use a dedicated service account with the minimum required permissions. Do not use a cluster admin kubeconfig.
- Rotate the token regularly. The
kubectl create tokencommand accepts a--durationflag. Set a rotation schedule that matches your security policy. - The kubeconfig grants write access to Functions in the namespace. Treat it as a sensitive credential.