Deployment of the ECS task definitions with Terraform

You can use Terraform to deploy the ECS task definitions needed for the Dask cluster.

Prerequisites

You need to have Terraform installed and configured with your AWS credentials.

Steps

1. Clone the LabCas workflows repository:

   git clone https://github.com/jpl-labcas/workflows.git
   

2. Create an environment directory from the provided template:

   cp -r workflows/terraform/environments/env_template workflows/terraform/environments/your_env_name
   

3. Edit the ml_cluster.tfvars file in your new environment directory to set your AWS region, VPC ID, subnet IDs, ECS cluster name, task CPU and memory, and desired count of tasks:

   region           = "your-aws-region"
   vpc_id           = "your-vpc-id"
   subnet_ids       = ["subnet-id-1", "subnet-id-2"]
   ecs_cluster_name = "your-ecs-cluster-name"
   task_cpu         = "512"
   task_memory      = "1024"
   desired_count    = 2
   

4. Edit the terraform.tfvars file in the Terraform directory specify the backend configuration for storing the Terraform state.

5. Navigate to the Terraform directory for ECS task definitions:

   cd workflows/terraform/application/ml_cluster
   

6. Initialize Terraform:

   terraform init -backend-config=../../environments/your_env_name/terraform.tfvars
   

7. Review the Terraform plan:

   terraform plan -var-file=../../environments/your_env_name/ml_cluster.tfvars
   

8. Apply the Terraform configuration to create the ECS task definitions:

   terraform apply -var-file=../../environments/your_env_name/ml_cluster.tfvars
   

9. Confirm the apply action when prompted.