brain_mri/infra/infra.md

MLflow Infrastructure with Terraform

This Terraform setup provisions the necessary AWS resources for running an MLflow Tracking Server, including:

• An EC2 Ubuntu instance
• An RDS PostgreSQL database
• An S3 bucket for storing MLflow artifacts
• A VPC with public subnets and internet access

Prerequisites

• Terraform installed
• AWS CLI configured (aws configure)
• An EC2 key pair created in AWS Console
• A terraform.tfvars file with the required variables

Required terraform.tfvars Example

aws_region           = "us-east-1"
aws_profile          = "default" or your profile name
ec2_key_name         = "mlfkp"
db_password          = "StrongPassword123"
artifact_bucket_name = "your-unique-bucket-name"

Note: Ensure your AWS credentials are in ~/.aws/credentials and properly configured using the AWS CLI:

aws configure

Commands

Initialize Terraform

terraform init

Validate configuration

terraform validate

Preview plan

terraform plan -var-file="terraform.tfvars"

Apply infrastructure

terraform apply -var-file="terraform.tfvars"

Destroy infrastructure

terraform destroy -var-file="terraform.tfvars"