Overview

Kubernetes Engine (GKE) is a managed, production-ready environment for deploying containerized applications.

This quickstart will guide you through the steps needed to set up a GKE cluster with Banzai Cloud Pipeline.

Prerequisites

  • Google credentials json file
  • Banzai CLI tool authenticated against the Pipeline instance

Create a Google secret

In order to access resources on Google Cloud, the appropriate credentials need to be registered in the Banzai Cloud Pipeline’s secret store. (The reference to this secret will be used later on instead of passing the credentials around)

Follow this guide to create GKE credentials

The values needed for the secret are available in the google credential json file. Issue the command below to register your secret in Pipeline (replace the content with the values from your google credentials json!)

banzai secret create <<EOF 
{
  "name": "google-secret",
  "type": "google",
  "values": {
    "type": "service_account",
    "project_id": "{{google-project-id}}",
    "private_key_id": "{{private-key-id}}",
    "private_key": "{{private-key}}",
    "client_email": "{{client-email}}",
    "client_id": "{{client-id}}",
    "auth_uri": "{{auth-uri}}",
    "token_uri": "{{token-uri}}",
    "auth_provider_x509_cert_url": "{{auth-provider-x509-cert-url}}",
    "client_x509_cert_url": "{{client-x509-cert-url}}"
  }
}
EOF
Id                                                                Name     Type    UpdatedBy  Tags
b32343e28d37e09c26d91b4271eaa8dd689b16d9f1aba07fdc73af2a27750309  google-secret  google  lpuskas    []

Create GKE clusters

Use the secret ID from the previous command!

You can check the available regions and instance types at our Cloudinfo service.

Create a 3 node GKE cluster

The command below creates a 3 node GKE cluster (that the master node is provided by the GKE service)

banzai cluster create <<EOF
{
  "name": "gke-cluster",
  "location": "us-east1-b",
  "cloud": "google",
  "secretName": "google-secret",
  "properties": {
    "gke": {
      "master": {
        "version": "{{gke-version}}",
      },
      "nodeVersion": "{{gke-version}}",
      "nodePools": {
        "pool1": {
          "count": 3,
          "autoscaling": true,
          "instanceType": "n1-standard-2",
          "preemptible": false,
          "minCount": 3,
          "maxCount": 4
        }
      },
      "projectId": "{{google-project-id}}"
    }
  }
}
EOF

Notes:

  • placeholders need to be replaced with values from your credentials.json file
  • supported versions can be retrieved from the Banzai Cloud Cloudinfo service:
curl https://banzaicloud.com/cloudinfo/api/v1/providers/google/services/gke/regions/us-west2/versions
  • all nodes forming the cluster will be of the same instance type (n1-standard-2)
  • as autoscaling is turned on, it might happen that on higher loads there will be 4 nodes in the cluster

Create a multi node GKE cluster — multiple instance types

The following example creates a multinode GKE cluster where the cluster contains multiple instance types (the nodePools property has multiple elements)

banzai cluster create <<EOF
{
  "name": "gke-cluster",
  "location": "us-east1-b",
  "cloud": "google",
  "secretName": "google-secret",
  "properties": {
    "gke": {
      "master": {
        "version": "{{gke-version}}",
      },
      "nodeVersion": "{{gke-version}}",
      "nodePools": {
        "pool1": {
          "count": 3,
          "autoscaling": true,
          "instanceType": "n1-standard-2",
          "preemptible": false,
          "minCount": 3,
          "maxCount": 4
        },
        "pool2": {
          "count": 1,
          "autoscaling": true,
          "instanceType": "m1-megamem-96",
          "preemptible": true,
          "minCount": 1,
          "maxCount": 2
        }
      },
      "projectId": "{{google-project-id}}"
    }
  }
}
EOF

Notable differences compared to the first example:

  • the cluster will be formed of nodes of different instance types (there are 2 nodepools defined)
  • the number of instances is controlled on a per nodepool basis
  • the second nodepool will be filled with preemptible instances

Check the status of the cluster

You can check the status of the cluster creation with the following command:

banzai cluster get "gke-cluster"

Once the cluster is ready, you can try it with some simple commands. banzai cluster shell executes a shell within the context of the selected cluster. If you type a command in the shell opened, or pass it as arguments, it will be executed in a prepared environment. For example, you can list the nodes of the cluster using the original kubectl command:

banzai cluster shell --cluster-name "gke-cluster" -- kubectl get nodes

Further steps

If you are happy with the results, go on with the Deploying workload guide to learn about the basic features of a cluster.