Part 2: Provision a Nutanix cluster¶

Now that our local environment is set up, we are ready to use Terraform to provision our Nutanix cluster. The module we are using provides some default behavior that we recommend following for this workshop. It creates a new Equinix Metal project for your cluster and a new VLAN in that cluster for private networking. It deploys 3 Nutanix nodes with private IPs and 1 bastion server that has a public IP address and can be used to access the Nutanix nodes via SSH. The Nutanix nodes are deployed to on-demand m3.large.x86 instances. These defaults--among other behaviors--can be changed depending on your workshop needs; we will discuss those options later in the workshop.

Steps¶

1. Set up your Terraform variables¶

In order to provision your cluster with Terraform, you must set some variables that the Terraform module will use to determine how and where you cluster is deployed. You can see examples of all of the variables supported in the module in terraform.tfvars.example.

For this workshop, we want to use the default values of most variables so we only need to specify values for the following variables:

metal_auth_token so that Terraform can create resources in Equinix Metal on your behalf
metal_project_name so that it is easy to find your Nutanix cluster in the Metal console or CLI
metal_metro so that Terraform knows where to deploy your infrastructure. We recommend deploying to the Seoul metro (sl) for the best experience.

Create a terraform.tfvars file that sets the variables mentioned above. Your terraform.tfvars should look something like this:

Terraform

metal_auth_token      = "<your Equinix Metal API token>"
metal_metro           = "sl"
metal_project_name    = "<your name>-nutanix-workshop"

2. Provision your infrastructure with Terraform¶

Now that we have configured the necessary Terraform variables, we are ready to provision our Nutanix cluster. Terraform will prompt us to review the planned deployment before proceeding:

Bash

$ terraform apply
# ...
# ... terraform plan output (we should probably show at least some of this?)
Do you want to perform these actions?
  Terraform will perform the actions described above.
  Only 'yes' will be accepted to approve.

  Enter a value:

Review the terraform plan to confirm that you see the expected number of Nutanix nodes, that all infrastructure is being deployed to the expected metro, etc., and then enter yes to proceed with the deployment.

Each Nutanix node will take about 20 minutes to deploy. All 3 nodes are deployed in parallel. Once all 3 nodes are deployed successfully, there is additional cluster configuration that takes about 20 minutes as well, so you should have a working Nutanix cluster in about 40 minutes.

If any of your Nutanix nodes fails to provision, wait for Terraform to finish running and then you can re-run terraform apply to have Terraform attempt to replace the failed nodes; keep in mind that any provisioning failures will extend the time it takes to get a working Nutanix cluster.

3. Verify¶

Once Terraform has finished running, you can verify that your Nutanix cluster has been provisioned by checking the Equinix Metal console. You should see a new project with the name you specified in your terraform.tfvars file. Inside that project, you should see a new VLAN and 4 instances: 3 Nutanix nodes and 1 bastion server. Additionally, you can connect to the bastion host to ensure it's up and available. The Terraform module provides several outputs you can use to connect to the bastion host. Run the following command to connect to the bastion host:

Bash

ssh -i $(terraform output -raw ssh_private_key) root@$(terraform output -raw bastion_public_ip)

Discussion¶

Before proceeding to the next part let's take a few minutes to discuss what we did. Here are some questions to start the discussion.

How can you reduce the possibility of provisioning failures for Nutanix nodes?
Can you change the version of Nutanix that is deployed?