Model My Watershed Tiler Deployment

• Development
• Pre-deploy setup
• Deploying
  • Via GitHub Actions
  • Manual Deploy
• Destroy
  • Via GitHub Actions
  • Manual Deploy

This repository deploys TiTiler (as the fork titiler-mosaicjson) and FilmDrop UI via the FilmDrop Infrastructure Terraform Modules.

Development

Install pre-commit hooks:

pre-commit install

And run them with:

pre-commit run --all-files

Due to an issue with VSCode, the wrong JSON Schema is selected for the file .github/workflows/deploy.yaml. To prevent this, add the following to your .vscode/settings.json:

{
    "yaml.schemas": {
        "https://json.schemastore.org/github-workflow.json": [".github/workflows/*.{yml,yaml}"]
    }
}

Pre-deploy setup

1. In the AWS Accounts to be deployed into, create the bootstrap resources as outlined in <bootstrap/README.md>.

Deploying

Via GitHub Actions

Create a GitHub Environment (e.g., staging) with these Environment Secrets:

Variable	Description	Example
AWS_REGION	region to deploy into	us-west-2
AWS_ROLE	the ARN of the AWS Role to use for deploy	arn:aws:iam::0123456789:role/appFilmDropDeployRoleBootstrap-DeployRole-Wfx5HwlneOVM
TF_STATE_BUCKET	the bucket use for storing Terraform state	filmdrop-{project_name}-{region}-terraform-state-{random_string}
TF_STATE_LOCK_TABLE	the DynamoDB table to use for Terraform locks	filmdrop-terraform-state-locks
SLACK_CHANNEL_ID	ID of Slack channel to post deploy status notifications	D26F29X7OB3
SLACK_BOT_TOKEN	Slack Bot Token	alphanumeric string

The following GitHub Actions will run under the following situations:

• The validation workflow will run upon any push to any branch. This runs some tests on the validity of the Terraform configuration.
• The staging workflow will run upon push to main or any tag starting with v.
• The prod workflow will run upon push to any tag starting with v.

The staging and prod workflows require manual approval to access their respective GitHub Environments.

Manual Deploy

By default, Terraform will use a local store for state. If you want to configure this to use S3 and DynamoDB instead, in the same way the GitHub Actions build does, create a file to define the backend named config.s3.backend.tf with contents like:

terraform {
  backend "s3" {
    encrypt        = true
    bucket         = REPLACE_ME # with the bootstrapped bucket name
    dynamodb_table = "filmdrop-terraform-state-locks"
    key            = "mmw-{username}-test.tfstate" # replace with username
    region         = "us-west-2"
  }
}

The bucket name will be the value to be set for TF_STATE_BUCKET, e.g., filmdrop-{project_name}-{region}-terraform-state-{random_string}.

Download the filmdrop-aws-tf-modules source:

./scripts/retrieve_tf_modules.sh v2.27.0

Re-run this anytime you with to uptake a new filmdrop-aws-tf-modules release, in addition to updating the env var in .github/workflows/ci.yaml.

Run the Terraform commands to initialize, validate, plan, and apply the configuration:

terraform init
terraform validate
terraform plan -var-file=staging.tfvars -out tfplan
terraform apply -input=false tfplan

If you prefer to use a local state file, can delete the config.s3.backend.tf file and run terraform init again without it.

Destroy

Via GitHub Actions

Not currently supported.

Manual Deploy

Run terraform destroy -var-file=staging.tfvars -input=false