website/content/blog/autodeployterraform.md

180 lines
3.7 KiB
Markdown
Raw Normal View History

2020-05-08 23:13:36 -04:00
---
title: "Automatic Deployments with Terraform"
date: 2020-05-08T22:45:18-04:00
draft: false
2022-11-12 11:22:54 -05:00
tags: ["Deployment"]
2020-05-08 23:13:36 -04:00
---
I have recently written about [Packer](https://brandonrozek.com/blog/snapshotswithpacker/) to create system images or snapshots. This post will go over another [HashiCorp](https://www.hashicorp.com/) project named [Terraform](https://www.terraform.io/) that we can use to deploy that image to a VPS. Like before, I am going to go over how to setup this up in DigitalOcean. Check out [this list](https://www.terraform.io/docs/providers/index.html) for documentation on your favorite cloud provider.
## Variables
To protect against committing secrets like API keys, we're going to create a file that only stores variables. For it to be loaded automatically, it needs to be named `terraform.tfvars`. Here is an example configuration:
```
region = "nyc3"
size = "512mb"
domain = "example.com"
subdomain = "temp"
# Secrets
do_token = "DO-TOKEN-HERE"
key_name = "SSH-KEY-NAME-ON-DO"
```
Now to define the variables in HCL, we need to create a separate `variables.tf` file defining their types
```
variable "do_token" {
type = string
}
variable "domain" {
type = string
}
variable "key_name" {
type = string
}
variable "subdomain" {
type = string
}
variable "region" {
type = string
}
variable "size" {
type = string
}
```
## Configuration
Now let's create a file called `do.tf`. We need to start off by stating which provider we are using
```
provider "digitalocean" {
token = var.do_token
}
```
If you want to hook up your SSH key, then we need to query the Digital Ocean API for its ID.
```
data "digitalocean_ssh_key" laptop {
name = var.key_name
}
```
We need to also query the API for the packer snapshot we created. Replace this with any standard image like `"ubuntu-20-04-x64"` if you don't want to use a snapshot.
```
data "digitalocean_droplet_snapshot" "packer_snapshot" {
name = "packer-example"
most_recent = true
}
```
Now we can create the droplet
```
resource "digitalocean_droplet" "web" {
name = "tf-1"
image = data.digitalocean_droplet_snapshot.packer_snapshot.id
region = var.region
size = var.size
ssh_keys = [data.digitalocean_ssh_key.laptop.id]
backups = false
}
```
Attach a domain to the droplet
```
resource "digitalocean_record" "www" {
domain = var.domain
type = "A"
name = var.subdomain
value = digitalocean_droplet.web.ipv4_address
}
```
Output useful pieces of information like the new system's IP address and the domain
```
output "ip" {
value = digitalocean_droplet.web.ipv4_address
}
output "domain" {
value = "${digitalocean_record.www.name}.${digitalocean_record.www.domain}"
}
```
The whole configuration file for your convenience:
```
provider "digitalocean" {
token = var.do_token
}
data "digitalocean_ssh_key" laptop {
name = var.key_name
}
data "digitalocean_droplet_snapshot" "packer_snapshot" {
name = "packer-example"
most_recent = true
}
# Create a droplet
resource "digitalocean_droplet" "web" {
name = "tf-1"
image = data.digitalocean_droplet_snapshot.packer_snapshot.id
region = var.region
size = var.size
ssh_keys = [data.digitalocean_ssh_key.laptop.id]
backups = false
}
# Attach a subdomain
resource "digitalocean_record" "www" {
domain = var.domain
type = "A"
name = var.subdomain
value = digitalocean_droplet.web.ipv4_address
}
output "ip" {
value = digitalocean_droplet.web.ipv4_address
}
output "domain" {
value = "${digitalocean_record.www.name}.${digitalocean_record.www.domain}"
}
```
# Deploy
Check if your configuration is valid
```bash
terraform plan
```
Deploy!
```bash
terraform apply
```
Take down when done
```bash
terraform destroy
```