Using Terraform

leafcloud/tutorials

In this tutorial I'm going to show an example of how to setup a highly available service on Leafcloud, using Terraform.

Configure the CLI if you have not done so already.

Login to Leafcloud, go to "identity", create and download "application credentials" as clouds.yaml. Now copy or move this file to ~/.config/openstack/clouds.yaml

Finally; tell the Openstack client to use your cloud:

export OS_CLOUD=openstack

I recommend putting this in your .bash_rc (linux) or .profile (mac) file, so it sets this in your shell on launch.

If you have not used the OpenStack CLI, this is also a good moment to install that. Installing is as simple as

pip install python-openstackclient

Then try:

openstack image list

If it returns a list of available server images your setup was correct.

Setup Terraform with the OpenStack driver

Terraform will read all files with .tf a extension, but if you follow this guide your folder will look something like this:

main.tf
networks.tf
instances.tf
loadbalancer.tf

main.tf

terraform {
required_version = ">= 0.14.0"
  required_providers {
    openstack = {
      source  = "terraform-provider-openstack/openstack"
      version = "~> 1.39.0"
    }
  }
}

# Configure the OpenStack Provider to use the (newer) Octavia loadbalancer
provider "openstack" {
  use_octavia = true
}

It's good to now run the setup. Tun terraform init. Terraform should now download the OpenStack driver and configure the relevant versions.

Create a private network and a Compute instance

Next, we're going to setup a (private) network, and a subnet.

networks.tf

resource "openstack_networking_network_v2" "private" {
  name = "private"
  admin_state_up = "true"
}

resource "openstack_networking_subnet_v2" "default_sn" {
  name = "default_sn"
  cidr = "192.168.42.0/24"
  ip_version = 4
  network_id = openstack_networking_network_v2.private.id
}

This would be a good time to do your first apply, to make it happen. Run terraform apply. Terraform should return you the plan it wants to execute. Enter 'yes' to apply. If no errors occur you should have newly network.

You can check with openstack network list

Define the instances you want to start

And now setup the actual compute instances. The following snippet will configure three instances to start.

Notice the "count = 3". This is a variable to indicate we want to have three of the same servers.

instances.tf

resource "openstack_compute_instance_v2" "blue" {
  # The amount of instances you want to start
  count = 3

  name            = "blue_${count.index}"
  flavor_name     = "ec1.large"
  key_pair        = "<your_key_name>"
  security_groups = ["default"]
  # user_data       = file("cloud-init-blue.yaml")

  network {
    name        = data.openstack_networking_network_v2.private.name
    # Here we get the value of the count and add 10, 
    # so that each instance gets it's own unique I
    fixed_ip_v4 = "192.168.42.${count.index + 10}"
  }
}

Now run tf apply again. Check with openstack server list that your servers were created.

Using LoadBalancers

The next part is to create a LoadBalancer. These are great for splitting the load over the different servers you have created.

Note that here, we use the 'count' variable on the members, so that each instance will be added as a member to the default pool.

loadbalancers.tf

resource "openstack_lb_loadbalancer_v2" "demo_lb" {
  name           = "demo-loadbalancer"
  vip_network_id = data.openstack_networking_network_v2.external.id
}

resource "openstack_lb_listener_v2" "listener_1" {
  protocol        = "HTTP"
  protocol_port   = 80
  loadbalancer_id = openstack_lb_loadbalancer_v2.demo_lb.id

  insert_headers = {
    X-Forwarded-For = "true"
  }
}

resource "openstack_lb_pool_v2" "main_pool" {
  protocol    = "HTTP"
  lb_method   = "ROUND_ROBIN"
  listener_id = openstack_lb_listener_v2.listener_1.id
}

resource "openstack_lb_members_v2" "members_1" {
  count   = 3
  pool_id = openstack_lb_pool_v2.main_pool.id

  member {
    address       = "192.168.42.${count.index + 10}"
    protocol_port = 80
  }
}

Now check that these are started: openstack loadbalancer list

If the Loadbalancer command is not found you can install it with python-octaviaclient. It will be installed as a plugin to your openstack CLI