These are Terraform modules and 'deployments' (i.e. compositions of modules and environment-specific configuration to define a given subset of our infrastructure) for the systems and services run by the SRCF.
terraform plan, terraform apply etc. should be run from within a given
deployment folder.
The Terraform state (which stores Terraform's memory of all resources it
manages between invocations of the tool) is stored and locked via a Postgres
database, accessible to the sysadmins. Each deployment has its own state, kept
in its own database schema. As a one-time action, when first running
terraform init the correct backend config argument must be passed to set up
this database as Terraform's state backend.
This is a work in progress. Currently, the aim of this Terraform code is to set up infrastructure to a minimal 'configurable' state, at which point our Ansible roles and playbooks take over. The Terraform code should not try to set things up any more, except where there is a sensible universal requirement (for example, installing the Xen guest agent on Xen VMs).