Provision an Unmanaged Kubernetes Cluster using Terraform

Terraform, the orchestration tool by HashiCorp, can be used to deploy a Kubernetes cluster on Linode. Linode’s Terraform K8s module creates a Kubernetes(K8s) cluster running on Ubuntu, and simplifies many of the steps involved in deploying a Kubernetes cluster with kubeadm. After creating master and worker nodes, the module connects over SSH to these instances and installs kubeadm, kubectl, and other Kubernetes binaries to the /opt/bin directory. It also initializes kubeadm, joins the worker nodes to the master, and configures kubectl to control the cluster. Calico is installed for the container networking interface of the cluster. A kubectl config file is installed to the local environment which connects to the API server of the cluster.

Note

Development work on the module is active. For the latest updates and validated Terraform configurations, see the module’s GitHub repository.

Before starting to deploy a Kubernetes cluster with Terraform, make sure:

1. You are familiar with Terraform. You can read through A Beginner’s Guide to Terraform to familiarize yourself with key concepts.

2. You are familiar with Kubernetes concepts. For an introduction, see the A Beginner’s Guide to Kubernetes series. Read through Getting Started with Kubernetes: Use kubeadm to Deploy a Cluster on Linode to get familiar with kubeadm.

3. You have a personal access token for Linode’s v4 API to use with Terraform. Follow the Getting Started with the Linode API to get a token.

   Note

   When creating a personal access token, ensure it is set to Read/Write access as new Linode servers are being created.

4. Terraform is installed on your computer. See Install Terraform for more information.

   Note

   This guide was written using Terraform version 0.12.24. The module requires at least Terraform 0.10.

5. And, lastly, kubectl is installed on your computer. The kubectl is necessary to connect to and manage the Kubernetes cluster; deployment using the Terraform module fails if kubectl is not installed locally. See Install kubectl for more information.

Deploying a Kubernetes cluster with Linode’s K8s Terraform module requires:

• a local environment with a kubectl instance
• a system-wide installation of Python
• SSH keys, SSH keys configured with the SSH agent
• the sed and scp command-line utilities The module’s script preflight.sh verifies these requirements are installed on the local environment and generates a $var not found error if any of the tools are missing. In this section learn how to:
• install and configure kubectl
• set up the SSH agent
• create an environment variable to store the API v4 token

If there is an error stating the system is missing Python, scp, or sed, use the operating system’s package manager to install the missing utilities.

Create a Python Alias

If Python is invoked using python3, alias the command so Terraform can execute scripts locally using Python as its interpreter. Using a text editor, edit ~/.bashrc file to include the following alias:

alias python=python3

Then, reinitialize ~/.bashrc file for the changes to take effect.

source ~/.bashrc

macOS:

Install via Homebrew:

brew install kubectl

If you don’t have Homebrew installed, visit the Homebrew home page for instructions. Alternatively, you can manually install the binary; visit the Kubernetes documentation for instructions.

Linux:

1. Download the latest kubectl release:

   curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
   

2. Make the downloaded file executable:

   chmod +x ./kubectl
   

3. Move the command into your PATH:

   sudo mv ./kubectl /usr/local/bin/kubectl
   

Note

You can also install kubectl via your package manager; visit the Kubernetes documentation for instructions.

Windows:

Visit the Kubernetes documentation for a link to the most recent Windows release.

By default, Terraform uses the SSH agent of the operating system to connect to a Linode instance through SSH. In this section learn how to run the SSH agent and add the necessary SSH keys to it.

1. Run the SSH agent with the following command:

    eval `ssh-agent`
   

   The output is similar to:

   Agent pid 11308
      

2. Add the SSH keys to the agent. For more information, see creating an authentication key-pair. This command adds keys from the default location, ~/.ssh/

    ssh-add
   

   The output is similar to:

   Identity added: /home/example_user/.ssh/id_rsa (/home/example_user/.ssh/id_rsa)
      

Before running the project, create an access token for Terraform to connect to the Linode API. Using the token and your access key, create the LINODE_TOKEN environment variable:

1
2

read -sp "Linode Token: " TF_VAR_linode_token # Enter your Linode Token (it will be hidden)
export TF_VAR_linode_token

This variable needs to be supplied to every Terraform apply, plan, and destroy command using -var linode_token=$LINODE_TOKEN unless a terraform.tfvars file is created with this secret token.

1. In the directory where Terraform was installed, create a new directory to store the configuration files of the K8s cluster.

    cd terraform
    mkdir k8s-cluster
   

2. Using a text editor, create the main configuration file of the cluster and name it main.tf. Add the following contents to the file.

   File: ~/terraform/k8s-cluster/main.tf

   1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

   terraform { required_providers { linode = { source = "linode/linode" version = "1.16.0" } } } module "k8s" { source = "linode/k8s/linode" version = "0.1.2" linode_token = var.linode_token server_type_master = var.server_type_master server_type_node = var.server_type_node cluster_name = var.cluster_name k8s_version = var.k8s_version region = var.region nodes = var.nodes }

   This file contains the main configuration arguments of the cluster. The only required configurations are source and linode_token. source calls Linode’s k8s module, while the linode_token gives access to viewing, creating, and destroying Linode resources.

   The rest of the configurations are optional and have some default values. In this example, however, Terraform’s input variables are used, so the main.tf configuration can be reused across different clusters.

3. Create an input variables file, named variables.tf, with the example content.

   File: ~/terraform/k8s-cluster/variables.tf

   1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

   variable "linode_token" { description = " Linode API token" } variable "server_type_master" { default = "g6-standard-2" description = " Linode API token" } variable "cluster_name" { description = " Linode API token" default = "example-cluster-1" } variable "server_type_node" { description = " Linode API token" default = "g6-standard-1" } variable "k8s_version" { description = " Linode API token" default = "v1.14.0" } variable "region" { description = "Values: us-east, ap-west, etc." default = "us-east" } variable "nodes" { description = " Linode API token" default = 3 }

   The example file creates input variables referenced in the main configuration file that was created. The values for those variables are assigned in a separate file in the next step. The default values of the k8s module can be overridden, as in the example file. For more details about input variables, see the Input Variables section in the A Beginner’s Guide to Terraform.

4. Create an input variables values file to provide the main configuration file with values that differ from the defaults in input variable file.

   File: ~/terraform/k8s-cluster/terraform.tfvars

   1 2 3

   server_type_master = "g6-standard-4" cluster_name = "example-cluster-2"

   In this example, the master node of the cluster uses a g6-standard-4 Linode plan, instead of the default g6-standard-2, and the cluster_name is set to example-cluster-2, instead of example-cluster-1.

1. Change to the ~/terraform/k8s-cluster/ directory and initialize Terraform to install the Linode K8s module.

    terraform init
   

2. Verify Terraform creates the resources of the cluster as expected before making any actual changes to the infrastructure. To do this, run the plan command:

    terraform plan
   

   This command generates a report detailing what actions Terraform takes to set up the Kubernetes cluster.

3. If satisfied with the generated report, run the apply command to create the Kubernetes cluster. This command prompts confirmation to proceed.

    terraform apply -var-file="terraform.tfvars"
   

   After a few minutes, when Terraform has finished applying the configuration, it displays a report of the actions were taken. And the Kubernetes cluster is ready for you to connect to it.

After Terraform finishes deploying the Kubernetes cluster, the ~/terraform/k8s-cluster/ directory should contain a file named default.conf. This file contains the kubeconfig file. Use kubectl, along with this file, to gain access to the Kubernetes cluster.

1. Save the path of the kubeconfig file to the $KUBECONFIG environment variable. In the example command, the kubeconfig file is located in the Terraform directory created at the beginning of this guide. Ensure the command is updated with the location of the default.conf file:

    export KUBECONFIG=~/terraform/k8s-cluster/default.conf
   

   Note

   It is common practice to store kubeconfig files in ~/.kube directory. By default, kubectl searches for a kubeconfig file named config that is located in the ~/.kube directory. Other kubeconfig files can also be specified by setting the $KUBECONFIG environment variable.

2. View the nodes in the cluster using kubectl.

    kubectl get nodes
   

   Note

   If the kubectl commands are not returning the resources and information you expect, then the client may be assigned to the wrong cluster context. Visit the Troubleshooting Kubernetes guide to learn how to switch cluster contexts.

You are now ready to manage the cluster using kubectl. For more information about using kubectl, see the Kubernetes Overview of kubectl.

A new terminal window does not have access to the context specified using the previous instructions. This context information can be made persistent between new terminals by setting the KUBECONFIG environment variable in the configuration file of the shell.

Note

If you are using Windows, review the official Kubernetes documentation to persist a context.

These instructions are for the Bash shell and they are similar to other shells:

1. Navigate to the $HOME/.kube directory:

   cd $HOME/.kube
   

2. Create a directory named configs within $HOME/.kube. This directory can be used to store the kubeconfig files.

   mkdir configs
   

3. Copy the default.conf file to the $HOME/.kube/configs directory.

    cp ~/terraform/k8s-cluster/default.conf $HOME/.kube/configs/default.conf
   

   Note

   Optionally, you can provide the copied file a different name to help distinguish it from other files in the configs directory.

4. Open up the Bash profile (~/.bashrc) in a text editor and add the configuration file to the $KUBECONFIG PATH variable.

   If an export KUBECONFIG line is already present in the file, append to the end of this line as follows; if it is not present, add this line to the end of the file:

   export KUBECONFIG=$KUBECONFIG:$HOME/.kube/config:$HOME/.kube/configs/default.conf
   

5. Close the terminal window and open a new window to receive the changes to the $KUBECONFIG variable.

6. Use the config get-contexts command for kubectl to view the available cluster contexts:

    kubectl config get-contexts
   

   The output should be similar to the following:

   CURRENT   NAME                                 CLUSTER             AUTHINFO           NAMESPACE
   *         kubernetes-admin@example-cluster-1   example-cluster-1   kubernetes-admin

7. If your context is not already selected, which is denoted by an asterisk in the current column, switch to this context using the config use-context command. Supply the full name of the cluster including the authorized user and the cluster:

   kubectl config use-context kubernetes-admin@example-cluster-1
   

   The output should be similar to the following:

   Switched to context "kubernetes-admin@example-cluster-1".

8. You are now ready to interact with the cluster using kubectl. Test the ability to interact with the cluster by retrieving a list of Pods. Use the get pods command with the -A flag to see all pods running across all namespaces:

   kubectl get pods -A
   

   The output should be similar to the following:

   NAMESPACE            NAME                                               READY   STATUS    RESTARTS   AGE
   kube-system          calico-node-5bkc6                                  2/2     Running   0          17m
   kube-system          calico-node-gp5ls                                  2/2     Running   0          17m
   kube-system          calico-node-grpnj                                  2/2     Running   0          17m
   kube-system          calico-node-qd85t                                  2/2     Running   0          17m
   kube-system          ccm-linode-mjgzz                                   1/1     Running   0          17m
   kube-system          coredns-fb8b8dccf-5tlbm                            1/1     Running   0          17m
   kube-system          coredns-fb8b8dccf-7tpgf                            1/1     Running   0          17m
   kube-system          csi-linode-controller-0                            3/3     Running   0          17m
   kube-system          csi-linode-node-gfd8m                              2/2     Running   0          17m
   kube-system          csi-linode-node-hrfnd                              2/2     Running   0          16m
   kube-system          csi-linode-node-q6fmd                              2/2     Running   0          17m
   kube-system          etcd-mytestcluster-master-1                        1/1     Running   0          16m
   kube-system          external-dns-7885f88564-tvpjf                      1/1     Running   0          17m
   kube-system          kube-apiserver-mytestcluster-master-1              1/1     Running   0          16m
   kube-system          kube-controller-manager-mytestcluster-master-1     1/1     Running   0          16m
   kube-system          kube-proxy-cs9tm                                   1/1     Running   0          17m
   kube-system          kube-proxy-qljn5                                   1/1     Running   0          17m
   kube-system          kube-proxy-sr5h8                                   1/1     Running   0          17m
   kube-system          kube-proxy-ww2tx                                   1/1     Running   0          17m
   kube-system          kube-scheduler-mytestcluster-master-1              1/1     Running   0          16m
   kube-system          kubernetes-dashboard-5f7b999d65-jk99z              1/1     Running   0          17m
   kube-system          metrics-server-58db9f9647-tz8f8                    1/1     Running   0          17m
   reboot-coordinator   container-linux-update-agent-6kgqm                 1/1     Running   0          16m
   reboot-coordinator   container-linux-update-agent-7nck5                 1/1     Running   0          17m
   reboot-coordinator   container-linux-update-agent-nhlxj                 1/1     Running   0          17m
   reboot-coordinator   container-linux-update-agent-vv8db                 1/1     Running   0          17m
   reboot-coordinator   container-linux-update-operator-5c9d67d4cf-78wbp   1/1     Running   0          17m

More Information

You may wish to consult the following resources for additional information on this topic. While these are provided in the hope that they will be useful, please note that we cannot vouch for the accuracy or timeliness of externally hosted materials.

• Kubernetes Terraform installer for Linode Instances