Skip to main content

GCP GKE Terraform

Deploy Datagrok on Google Cloud Platform using a Terraform module that provisions GKE, Cloud SQL (PostgreSQL), GCS, and a VPC, then installs the Helm chart into the cluster. This is the GCP analogue to the AWS CloudFormation (EKS) template.

The same Datagrok services run on every deployment — see Components for the canonical list.

Already have a GKE cluster?

If your team already manages GKE, install the Helm chart directly with the values-gke.yaml overlay (Cloud SQL + GCS + Workload Identity). The Terraform module is for stands that need GKE, Cloud SQL, GCS, IAM, and the chart provisioned together.

Prerequisites​

  • GCP project with billing enabled and the GKE, Cloud SQL, Cloud Storage, and Compute Engine APIs enabled.
  • Terraform 1.5+ and a GCP service account credential with Project Owner (or the equivalent narrower permissions: GKE admin, Cloud SQL admin, Storage admin, IAM admin, Network admin).
  • gcloud and kubectl installed locally.

Deploy​

The Datagrok GCP Terraform module creates the GKE cluster, Cloud SQL instance, GCS bucket, and supporting IAM, then installs the Helm chart into the cluster.

module "datagrok_gcp" {
  source  = "github.com/datagrok-ai/tf-module-datagrok-core//gcp"
  project = "<gcp-project-id>"
  region  = "europe-west3"

  cluster_name      = "datagrok"
  datagrok_version  = "1.27.3"
  datagrok_dns      = "datagrok.example.com"
  acm_cert          = ""        # use Google-managed cert via the chart's ingress instead
}

Apply:

terraform init
terraform apply

After Terraform finishes, the chart is already installed and Datagrok is reachable through the ingress. Configure kubectl to inspect the running stand:

gcloud container clusters get-credentials datagrok --region europe-west3 \
  --project <gcp-project-id>
kubectl -n datagrok get pods

Service versions​

The Terraform module's datagrok_version input pins all service image tags to the same release. Defaults track the latest stable Datagrok release; see the release history and the equivalent Helm chart overrides for granular per-service control.

Update Datagrok components​

Bump datagrok_version (and the related variables, if you pinned services individually) and re-apply:

terraform apply -var datagrok_version=1.27.4

Cloud SQL and the GCS bucket are not replaced on apply. Database schema migrations run automatically on the first start of the new app version. For a manual upgrade with the chart only (no Terraform changes), see Helm chart upgrades.

See also​