Skip to content

Latest commit

 

History

History
337 lines (221 loc) · 13.1 KB

week1.md

File metadata and controls

337 lines (221 loc) · 13.1 KB

Week 1 - Building Terrahouse

Table of Contents

Root Module

The root module structure shoud be:

PROJECT_ROOT
│
├── main.tf                 # everything else.
├── variables.tf            # stores the structure of input variables
├── terraform.tfvars        # the data of variables we want to load into our terraform project
├── providers.tf            # defined required providers and their configuration
├── outputs.tf              # stores our outputs
└── README.md               # required for root modules

Module Structure Guidelines

Variables in Terraform Cloud

We can store Terraform Variables (e.g. vars set in .tfvars file) or Environment Variables (e.g. vars set in bash terminal, AWS Credentials).

These can be set to sensitive to avoid exposure.

(IDE...) Loading Terraform Input Variables

Variable defaults > -var or-var-file > .autotfvars or .autotfvars.json > terraform.tfvars.json > terraform.tfvars > Environment variables (TF_VAR_variable_name)

variables.tf file (input variable documentation)

Example:

variable "image_id" {
  type        = string
  description = "The id of the machine image (AMI) to use for the server."
}

variable "availability_zone_names" {
  type    = list(string)
  default = ["eu-west-2"]
}

This is the format variables are specified, by name and type. Other options include descriptions and default values to be used if a variable isn't loading in.

var flag

We can use the var flag to set a input variable or override a variable in the tfvars file e.g. terraform -var user_uuid="my-user_uuid".

var-file flag

To set lots of variables, it is more convenient to specify their values in a variable definitions file (with a filename ending in either .tfvars or .tfvars.json) and then specify that file on the command line with -var-file.

.tfvars file

A variable definitions file uses the same basic syntax as Terraform language files, but consists only of variable name assignments.

.auto.tfvars has higher precedance than terraform.tfvars but acts the same.

terraform.tfvars is the defult file to load in many Terraform variables in one go.

Env Vars

A less prioritised way of defining variables for Terraform, is the use of env vars prefix with TF_VAR_ then the name of a declared variable.

Terraform Variables

Dealing With Configuration Drift

Lose of state file may have to be resolved with partial/complete teardown. However, depending on the resources, it can be rectified.

Fix Missing Resources With Terraform Import

Using terraform import we can import provisioned resources that are missing from the .tfstate file. By checking the provider documentation, we see what resources can be imported and the appropriate command syntax to do so.

terraform import aws_s3_bucket.bucket bucket-name

You can also do the same by using an import block, like in this example:

import {
  to = aws_s3_bucket.bucket
  id = "bucket-name"
}

Example for Importing S3 Bucket

Fix Manual Configuration

Resources have the potential to be manually deleted through ClickOps, the use of the AWS Console/GUI.

Here terraform plan does a good job of acknowledging the difference or drift, and attempts to reinstate infrastructure back to expected state.

Fix Using Terraform "Refresh"

This command reads the current settings from all managed remote objects an updates the Terraform state to match.

terraform apply -refresh-only --auto-approve

AWS Terrahouse Module

Modules allow for better organisation of infrastructure, into categories such as storage or content delivery. Where related resources a grouped together and enable reusability of resource configurations with Terraform.

Module Sources

This part of the module block allows Terraform to know where to source the configuration files for the desired child module, i.e terrahouse_aws:

module "terrahouse_aws" {
  source = "./modules/terrahouse_aws"
  user_uuid = var.user_uuid
  bucket_name = var.bucket_name
}

As shown above, input variables can be passed into the module, and these variables need to be declared in its own variables.tf file and more simply into the root module.

Using the source we can import the module from various places e.g:

  • Locally
  • Github
  • Terraform Registry

S3 Static Website Hosting

In this bootcamp we configure with Terraform, meaning provising infrastructure, configuration manangement (provisioners - not ideal, Ansible more production standard) and data configuration (of files). To learn the capabilities of Terraform.

Working With Files in Terraform

Fileexists function

This is a built in terraform function to check the existance of a file.

condition = fileexists(var.error_html_filepath) https://developer.hashicorp.com/terraform/language/functions/fileexists

Filemd5

This built in function hashes the contents of the file which we use to check content status of a file.

https://developer.hashicorp.com/terraform/language/functions/filemd5

Path Variable

In terraform there is a special variable called path that allows us to reference local paths:

path.module = get the path for the current module path.root = get the path for the root module

Special Path Variable

resource "aws_s3_object" "index_html" {
  bucket = aws_s3_bucket.website_bucket.bucket
  key = "index.html"
  source = "${path.root}/public/index.html" }

Typo Error with S3 bucket configuration:

During the Content Delivery Network lecture I found that the error.html file kept being served up as the default_root_object, which is set to index.html. After some investigation I realised the files had the same etag, so the same file must be duplicated for some reason or being uploaded as the source for both html objects.

Eventually I discovered that I accidently passed the wrong variable into the index_html_filepath:

index_html_filepath = var.error_html_filepath

I rectified this by change replacing the error with the correct matching variable index:

index_html_filepath = var.index_html_filepath

NOTE: It reminded me to be more careful to avoid typos and how easy it is to miss. Checking the contents of the S3 object in question earlier would have sped up the troubleshooting process.

Content Delivery Network Implementation

We are using CloudFront as our Content Delivery Network, which provides a globally-distributed network of proxy servers to cache content, e.g. html files and images, more locally to users which reduces latency and improves user experience.

Terraform Locals

Local variables can be defined uses locals. Useful when transforming data to another format and then referencing it as a variable.

locals {
  s3_origin_id = "MyS3Origin"
}

Local Values

Terraform Data Sources

Data Sources are used to get information from external/internal resources and insert that information into our Terraform Configuration.

AWS Data Sources

Setup Content Version

Changing the Lifecycle of resource

Meta-Arguments Lifecycle

Terraform_data

Plain data values such as Local Values and Input Variables don't have any side-effects to plan against and so they aren't valid in replace_triggered_by. You can use terraform_data's behavior of planning an action each time input changes to indirectly use a plain value to trigger replacement.

resource "aws_s3_object" "index_html" {
  #
  etag = filemd5(var.index_html_filepath)
  `lifecycle` {
    replace_triggered_by = [terraform_data.content_version.output]
    ignore_changes = [etag]
  }
}

Invalidate CloudFront Cache

Provisioners

Provisioners allow the execution of command on a remote terminal, configuring the commands in Terraform.

Typically used as last resort, with Terraform discourages its use, as software like ansible do a better job.

Declaring Provisioners

Local-exec

This will execute a command on the machine running the terraform commands e.g, plan apply

Remote-exec

This will execute commands on a remote terminal you target. Will need authentication to get into machines such as ssh.

Heredoc String

This programming principle allows you to express multiple-line string more clearly. Using backslash \ to divide string into seperate lines.

<<COMMAND
aws cloudfront create-invalidation \
--distribution-id ${aws_cloudfront_distribution.s3_distribution.id} \
--paths '/*'
    COMMAND

Heredoc Strings

Assets Upload

Here we created an assets folder to hold our images and whatever else we may decorate our house with.

For_Each Expressions

If a resource or module block has a for_each meta-argument, whose value is a map or a set of strings, Terraform creates one instance for each member of that map of set.

Additionally, we can use the fileset() function to target all files within a specified directory. E.g:

for_each = fileset("${path.root}/public/assets", "*.{jpg,png,gif,svg,mp3}")

Each.key

Where for_each is set, an additional each object can be used to pass an iteration from for_each to the each.key.

source = "${path.root}/public/assets/${each.key}"

Path.root

Note in the above code block, the ${path.root}, which is the filesystem path of the root module of the configuration. Helps make setting the paths a little easier, especially in regards to the Terraform configuration.

Cloudfront issue or js issue

Cloudfront distro makes me download the mp3 files and I was not sure how to make it play them instead.

  • The upside is that it randomly passes on an mp3 into the <embed> src attribute, so the shuffle/randomiser in random.js works. Don't think I want users to be able to toggle, as the intention was for a slightly different audio experience on each visit.
  • But on the local http-server, it doesnt change between the first to mp3 files in the audioFiles list. These are the first to mp3 files I added when tested code.

Fix: Apparently the <embed> tag element isn't as useful for audio in html5, with a better alternative being the <audio> tag. With the addition of controls and mute attributes, I could stop the automatic downloads and allow users the option of whether or not they want audio. Provided a better, less abrupt audio/user experiece.

Future Improvements: While trying to troubleshoot the audio issue, I came across the Web Audio API. I initially wanted custom volume controls, not using the default browser controls, so I will experiment with this to hopefully have a more bespoke look.