Terraform module for provisioning a Datafold dedicated-cloud deployment on GCP (GKE cluster + supporting infra).
Unlike the AWS module (terraform-aws-datafold, which creates a per-component IRSA role for server/worker/dfshell/dma/etc. in modules/eks/roles.tf), this repo has no equivalent per-component GCP service account module. modules/gke/iam.tf only creates a single generic node-pool service account with logging/monitoring/storage roles — there's no server/worker/dfshell/dma identity anywhere.
examples/deployment/infra/config.tf references module.gcp.server_role_arn, module.gcp.dfshell_service_account_name, etc. for every component — but module "gcp" doesn't exist anywhere in this repo. Treat that example as a dead/aspirational stub (likely copied from the AWS example and never finished), not a working pattern to copy.
Consequence: no current GCP dedicated-cloud customer uses Bedrock-equivalent (Vertex AI) or other cloud-identity-gated features — there's no infra to grant them yet. If a GCP customer needs one (e.g. Vertex AI for Thunderbolt/Memgraph), building the per-component Workload Identity SA module here is a prerequisite, not a config toggle. The Helm side (helm-charts) already has a dedicated serviceAccount.gcpServiceAccount field per component ready to consume a real SA email once this module exists.
No Vertex-AI-equivalent to AWS's k8s_access_bedrock variable exists (grep -rn "vertex\|bedrock\|llm" returns nothing). Would need to be added alongside the per-component SA module above.
Never reference customer names, deployment names, or account IDs in code, comments, docs, or examples here. Use the established generic placeholder convention instead (see examples/deployment, which uses acme-datafold as the example deployment name).