Skip to content

This repo will seamlessly setup self managed Kubernetes cluster in GCP using Terraform and Kubespray.

Notifications You must be signed in to change notification settings

vladwa/K8s-Cluster-Provisioner-GCP-Terrafrom

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

K8s-Cluster-Provisioner-GCP-Terrafrom

Provision a Kubernetes cluster on GCP using Terraform and Kubespray

Workflow

Overview:

This project will create:

  • Creation of Kubernetes master Instance template and Instance Group (1 & 2).The Compute instances are created in different zones.
  • Creation of Kubernetes etcd Instance template and Instance Group (3 & 4).The Compute instances are created in different zones.
  • Creation of Kubernetes worker nodes/minions Instance template and Instance Group (5 & 6).The Compute instances are created in different zones.
  • Creation of kubespray ansible Instance Template and Instane Group(7 & 8).
  • Installing the pre-requisite packages/modules/Scripts (pip,git,etc) required for kubespray(9).
  • Download the kubespray or git clone the kubespray project(9).
  • Coping the required scripts to generate hosts inventory file. Which will be provided as input to the kubespray cluster.yml playbook to setup kubernetes cluster(9).
  • Login to the kubespray ansible instance and execute the cluster.yml ansible-playbook to setup cluster. The cluster execution details will be shown in terminal and also the output is redirected to the “output” file(10).
  • Kubespray-Ansible Machine Ip and One of the Kube-master Ip will be displayed as part of output.

Status

This will install a Kubernetes cluster on GCP.

Approach

The terraform configuration inspects variables found in variables.tf to create resources in GCP.

The terraform script will take care of creating Master nodes,etcd nodes,worker nodes/minions, kubespray-ansible-node based on the configuration details.

There is a python script that generates a dynamic inventory that is consumed by the kubespray cluster.yml.

Kubernetes Nodes

You can create different kubernetes topologies by setting the below mentioned variable to indicate number of hosts.

  • Master nodes: kube_master_target_size variable

  • Etcd nodes : kube_etcd_target_size variable

  • Kubernetes worker nodes or minions: kube_minion_target_size variable

  • Kubespray ansible node: kube_ansible_target_size variable

Note that the kubespray Ansible script will report an invalid configuration if you wind up with an even number of etcd instances since that is not a valid configuration. Also it is recommended to have multiple master nodes for high availability.

Prerequisites

  • Install Terraform
  • Service account key json (Appropirate roles to be assinged to service account to create computes)
  • Centos or Redhat os to be used for kubespray ansible instance.
  • You have a pair of keys generated which is the part of image that can be used to secure the new hosts.
  • Kubespray. Make sure the project name is kubespray.

Configuration

Service Account key json file

Details from service account key json file should be updated in the account.json file The Project ID associated with the service account key should be set in the variables.tf

Note: To deploy several clusters within the same project you need to use terraform workspace.

SSH Key Setup

SSH keypair is required by Kubespray-Ansible to access the newly provisioned Instances on GCP.When ssh keys are generated for the user

Cluster variables:

The creation of the cluster is driven by values found in variables.tf or (cluster.tfvars)[./clustertffiles/cluster.tfvars]

For your cluster, edit clustertffiles/cluster.tfvars.

env variable is used to set a tag on each server deployed as part of this cluster.This helps with indentification of hosts associated with each cluster.

region variable is used to set in which region "Compute Instance templates and Compute Instance Groups" needs to be created.

gcp_project variable is used to set the GCP project_id.

user_name variable is used to set user name in kubespray inventory host file.

Ensure that username set for user_name matches the username used for SSH key generation.

kube_automation_folder the folder location where kubespray should be downloaded. The default value of this is '/home/app/kubespray', where app is the newly created user as part of SSH key setup.

kubespray_repo_url the kubespray git project url, user name and password to be passed in the URL. If required user credentials to be passed in the url like https://username:[email protected]/username/kubespray.git

kube_{component}_machine_type variable is used to set Compute Instance machine type.

kube_{component}_source_image variable is used to set OS Image type. This determines the operating system installed on the system.

kube_{component}_disk_size_gb variable is used to set Compute Instance disk size. Specifies the size of the disk in base-2 GB.

kube_{component}_disk_type variable is used to set Compute Instance disk type.

kube_{component}_network_interface variable is used set Network Interface.

kube_{component}_subnetworkvariable is used to set subnetwork.

kube_{component}_mode is used to set mode in which to attach this disk, either READ_WRITE or READ_ONLY.

kube_{component}_svca_email variable is used to set email address of the service account.

kube_{component}_svca_scopes variable is used to set list of scopes to be made available to the service account.

kube_{component}_target_size variable is used to set total number of instances in the group.

Where component can be either : master,etcd,minion or ansible

variables.tf or (cluster.tfvars)[./clustertffiles/cluster.tfvars] files are updated with default values.

Initialization

Before Terraform can operate on your cluster you need to install the required plugins. This is accomplished as follows:

$ cd clustertffiles
$ terraform init 

Provisioning cluster

You can apply the Terraform configuration to your cluster with the following command issued from your cluster's clustertffiles directory (cd clustertffiles):

$ terraform apply -var-file=cluster.tfvars

Destroying cluster

You can destroy your new cluster with the following command issued from the cluster's clustertffiles directory:

$ terraform destroy -var-file=cluster.tfvars

On executing above command all the Instance templates, Instance Groups and Instances related to that cluster will be deleted.

Please Note this action is irreversible.

Debugging

Enable debugging logs from Terraform by setting TF_LOG to DEBUG before "Provisioning cluster" step.

Kubernetes

Master Node access

Login to the kube master node with Ip displayed as part of output.

  • Execute below command in kube master node to verify the version details:
kubectl version
  • Verify that Kubernetes configuration file contains cluster details:
cat /root/.kube/config
  • Verify that all the nodes are up & running, using below command:
kubectl get nodes

What's next

Try out your new Kubernetes cluster with the Hello Kubernetes service.