Terraform Component

Terraform is a popular infrastruture as code technology often used to deploy cloud resources. We do provide our own opinionated way how to deploy Terraform. In this case you can follow simple conventions and you don't require to write a deployment scripts

Component Conventions

If component follows the conventions below, then hubctl will know how to deploy it.

Minimalistic terraform component will look like this:

Hubctl will automatically detect Terraform code when component contains one or more *.tf files.

./
├── hub-component.yaml  # component manifest
└── main.tf             # terraform code

Configuration

Terraform variables can be supplied in two ways:

Terraform variables can be defined by component parameter and exported as TF_VAR_* environment variable (recommended way)
Terraform variables can be defined in *.tfvars or *.tfvars.template file

Parameters

As every script, helm deployment scripts has been controlled via set off well-known environment variables. These variables should be defined in parameters section of hub-component.yaml. List of expected environment variables

Variable	Description	Required	Passed from `.env`
`COMPONENT_NAME`	Corresponds to parameter `hub.componentName` parameter, however can be overriden in `hub-component.yaml`. Hub will use this variable as a helm release name	x
`HUB_DOMAIN_NAME`	FQDN of a stack. We use this parameter as a natural id of the deployment	x	x
`HUB_CLOUD_PROVIDER`	Tells hubctl to use different backends for terraform. We currently support: `aws`, `azure` or `gcp`	x	x
`HUB_STATE_BUCKET`	Object storage bucket to be be used for terraform state	x	x
`HUB_STATE_REGION`	Region for terraform state bucket	x	x

There are additional environment variables, depends on your cloud type

AWS specific variables

These variables has been set during hubctl stack configure state and defined in .env file. You don't have to refer it for your component, however you can overwrite.

Variable	Description	Required
`AWS_PROFILE`	AWS name of the profile. Referenced from `.env` file. However you can override it in `hub-component.yaml` file	x

Azure specific variables

These variables has been set during hubctl stack configure state and defined in .env file. These are the minimum viable variables expected by our terraform deploymenet script

Variable	Description	Required
`ARM_CLIENT_ID`	The client(application) ID of an App Registration in the tenant
`ARM_CLIENT_SECRET`	A client secret that was generated for the App Registration
`ARM_SUBSCRIPTION_ID`	Access an Azure subscription value from within a Terraform script
`ARM_TENANT_ID`	ARM Tenant id

Full list of environment variables for azure can be found here

GCP specific variables

Variable	Description	Required
`GOOGLE_APPLICATION_CREDENTIALS`	Default applicaiton credentials (ADC) see details here
`GOOGLE_PROJECT`	For to refer google project ID. See details here

Full reference of supported variables available here

Deployment hooks

User can define pre and post deployment

pre-deploy to trigger action before helm install
post-deploy to trigter action after successful helm install
pre-undeploy to trigger action before helm delete
post-undeploy to trigger action after helm successful delete
import a special script that to import existing resources into Terraform state.

Note: import maybe a necessary step to allow manage non-idempotent resources such as AWS S3 bucket. Before running terraform deployment that will fail you may want to import bucket if it already exists. Note: deployment hooks should have execution rights