This tutorial covers the general creation of an Elastigroup, from an existing GKE cluster. The GKE import template is available under “Use Cases” in the Creation Wizard. The import wizard will assist you in importing all of the groups relevant settings directly from your Google Cloud account.
Step 1: General Settings
- Enter a name for your Elastigroup.
- Select either a Zonal or Regional GKE cluster under Location Type. Elastigroups supports both the default GKE Zonal clusters as well as Regional GKE clusters. To learn more about Regional GKE clusters see GKE’s documentation here.
- Select the zone from which you would like to import your GKE cluster.
- Choose an existing cluster and a node pool you would like to import.
Step 2: Configuration Settings
- Select the zones which you would like Elastigroup to consider when launching your instances. To maximize your savings Elastigroup calculates the cost of launching Preemptible VMs in the selected Zones. We recommend selecting multiple Zones to increase the number of Preemptible VMs markets available for Elastigroup to consider.
- Set the groups minimum, maximum, and target capacity.
- Select the instance types you’d like Elastigroup to consider when launching your instances. We recommend selecting multiple instance types to provide Elastigroup with as many Preemptible VMs market options as possible.
- Choose the OS image used by Elastigroup to launch the instances from.
Step 3: Auto Scaling (Optional)
- Configure whether you would like to use the Spotinst Kubernetes autoscaler. The autoscaler will automatically add or remove capacity from your cluster according to your overall node utilization and pod scheduling constraints. This ensures that you have enough capacity to run your workloads while maintaining ideal node utilization.
- Set your cluster’s headroom. Configuring headroom, allows you to tell the autoscaler to reserve CPU/Mem units for future pods to run on the cluster.
Step 4: Advanced Settings (Optional)
- Configure additional details such as the instances disk type and size, labels, and required metadata.
Step 5: Cluster Connectivity
- To get additional data about your GKE cluster in the Spotinst Console, you must run the Spotinst controller using the provided script. Run the script on any machine that is properly configured with kubectl to connect to your cluster.
- You can make sure the controller is running by executing the following command:
kubectl get pods -n kube-system
and make sure you see “spotinst–kubernetes-cluster-controller” running.
- You can refer to our Spotinst controller documentation page, for more helpNote: It might take up to 2 minutes for the spotinst controller to report back to the spotinst console .