The Kubernetes integration relies on using
kubectl's native authentication to connect to your cluster. You can use the
$KUBECONFIG environment variable to find the location of the Kubernetes configuration file, and configure any credentials required.
You should perform any custom authentication as part of a before init hook to make sure that
kubectl is configured correctly before any commands are run in after init and subsequent hooks.
The following sections provide examples of how to configure the integration manually, as well as using cloud provider-specific tooling.
Manual configuration allows you to connect to any Kubernetes cluster accessible by your Spacelift workers, regardless of whether your cluster is on-prem or hosted by a cloud provider. The Kubernetes integration automatically sets the
$KUBECONFIG environment variable to point at
/mnt/workspace/.kube/config, giving you a number of options:
- You can use a mounted file to mount a pre-prepared config file into your workspace at
- You can use a before init hook to create a kubeconfig file, or to download it from a trusted location.
Please refer to the Kubernetes documentation for more information on configuring kubectl.
The simplest way to connect to an AWS EKS cluster is using the AWS CLI tool. To do this, add the following before init hook to your Stack:
$CLUSTER_NAMEenvironment variables must be defined in your Stack's environment.
- This relies on either using the Spacelift AWS Integration, or ensuring that your workers have permission to access the EKS cluster.
The simplest way to connect to an AKS cluster in Azure is using the Azure CLI to automatically add credentials to your kubeconfig. To do this your stack needs to use a custom runner image with the Azure CLI and kubelogin installed, and needs to run some before init hooks to authenticate with your cluster. Depending on your exact use case, you may need to use slightly different commands. This guide outlines two main scenarios.
Please note that both examples assume that your stack has an
$AKS_RESOURCE_GROUP environment variable configured containing the name of the AKS cluster and the resource group name of the cluster respectively.
Using the Spacelift Azure Integration»
When using our Azure integration, you can use the computed
$ARM_* environment variables to login as the Service Principal for the integration:
1 2 3 4 5
Using private workers with Managed Identities»
When using private workers with a managed identity, you can use the identity of that worker to login:
1 2 3
You can use the gcloud CLI to authenticate with a GKE cluster when using the Spacelift GCP integration using the
gcloud container clusters get-credentials command. For this to work, you need to use a custom runner image that has the
gcloud CLI and
The Spacelift GCP integration automatically generates an access token for your GCP service account, and this token can be used for getting your cluster credentials as well as accessing the cluster. To do this, add the following before init hooks to your Stack:
1 2 3 4 5 6 7 8 9 10 11 12
Please note, your Stack needs to have the following environment variables set for this script to work:
GKE_CLUSTER_NAME- the name of your cluster.
GKE_CLUSTER_REGION- the region the cluster is deployed to.
GCP_PROJECT_NAME- the name of your GCP project.
get-credentials command configures your kubeconfig file to use the
gcloud config config-helper command to allow token refresh. Unfortunately this command will not work when we only have an access token available. The script provided works around this by manually removing and re-creating the user details in the config file.
Single Zone Deployment»
If your cluster is deployed to a single zone, you can use the
--zone flag instead of the
--region flag in the
gcloud container clusters get-credentials command:
1 2 3 4