Tear down the stack

Teardown destroys everything billable with a full tofu destroy of the main stack. It keeps the S3 state bucket, the ECR repos (cents), and the CI bootstrap role — so a later deploy brings the stack straight back.

Caution

This is destructive and irreversible. Postgres has no backups configured yet (see the roadmap), so tearing down loses all data. Only do this for the demo/idle lifecycle.

From CI (recommended)

Actions → Teardown → Run workflow → type destroy in the confirm box. Anything else aborts. See Deploy via CI.

Locally

export AWS_PROFILE=<sso-profile>
./scripts/teardown.sh

The script scales the fck-nat ASG to 0 and waits for the NAT instance to terminate before tofu destroy — otherwise the relaunching NAT instance holds the security group Tofu is trying to delete and the destroy deadlocks. It then runs tofu -chdir=infra destroy -auto-approve.

When it finishes, idle cost is back to ~€0 (S3 state + ECR only).