Our current architecture utilizes a single Terraform state file and workspace (monolith) across multiple distinct pipelines (e.g., separate pipelines for S3, VPC, and EC2) for an AWS account.

Since we had 19 different aws accounts, different state file for each resources was not feasible.

So currently all my resources pipeline updates the same tfvars based on target apply.

While we can create resources granularly, we currently lack the ability to safe-delete them. Running a standard terraform destroy in any single pipeline creates a plan to destroy all resources in the shared state file. So we need to update the CI/CD pipeline destroy logic to accept a --target argument (e.g., -target=module.s3_bucket).