Google Cloud

The script creates a dedicated VPC and subnet, static public IPv4, HTTP/HTTPS firewall rules, an IAP-only SSH rule and one Ubuntu 24.04 VM. It runs the full Compose stack on that VM. The default is e2-standard-4 with a 100 GB persistent boot disk. These are billable Google Cloud resources.

Before starting

Use a billing-enabled project. Your identity needs permission to enable the Compute API and create Compute Engine networking, disks and instances. For administrative access, use OS Admin Login and IAP tunnel permissions. The VM has no service account and no Google API scopes; it pulls public images without cloud credentials.

See Google's IAP TCP forwarding setup and OS Login setup.

Deploy from Cloud Shell or your workstation

git clone https://github.com/grovs-io/self-host.git
cd self-host
GROVS_RELEASE_URL="$(curl -fsSL -o /dev/null -w '%{url_effective}' https://github.com/grovs-io/self-host/releases/latest)"
./deploy/gcp/deploy.sh YOUR_PROJECT_ID europe-west1-b example.com [email protected] "${GROVS_RELEASE_URL##*/}"

The version lookup selects the latest completed release when you run the command. Both backend and dashboard use that same GHCR tag; see Images and releases.

The script shows the target project, zone and machine type before requesting confirmation. It passes --project explicitly and does not change your gcloud default project. Set GROVS_INSTANCE_NAME for another installation name and GROVS_MACHINE_TYPE for a different machine size.

For a repeatable install, set GROVS_STACK_REF to a reviewed self-host commit containing these deployment files. Application images use the release passed as the last argument; GROVS_STACK_REF selects the Compose files and setup scripts.

Point the DNS records at the printed static IP. Package installation and migrations continue after VM creation. Use the SSH command printed by the script, then:

sudo journalctl -u google-startup-scripts.service --no-pager -n 100
sudo systemctl status grovs --no-pager
sudo cat /opt/grovs/.env

Open https://dashboard.<your-domain> and sign in with the bootstrap admin values from .env. Then verify the deployment.

Restarts and upgrades

grovs.service starts the existing stack on reboot. Bootstrap preserves an existing checkout and .env; it does not regenerate secrets or change the application version. Follow the upgrade procedure from /opt/grovs using root access.

Failed or partial deployments

The script stops at the first failed cloud command. It does not automatically delete resources already created. Inspect the named resources before retrying; an existing name causes create to fail instead of replacing your installation. Use the commands below to remove a partial deployment, or finish provisioning manually. If VM startup failed, fix the reported issue before rerunning bootstrap.

Remove the deployment

Back up your data and .env first. Replace the values below if you chose a different project, name or location:

PROJECT=YOUR_PROJECT_ID
ZONE=europe-west1-b
REGION=europe-west1
NAME=grovs
gcloud --project="$PROJECT" compute instances delete "$NAME" --zone="$ZONE"
gcloud --project="$PROJECT" compute addresses delete "$NAME-ip" --region="$REGION"
gcloud --project="$PROJECT" compute firewall-rules delete "$NAME-web" "$NAME-ssh"
gcloud --project="$PROJECT" compute networks subnets delete "$NAME-subnet" --region="$REGION"
gcloud --project="$PROJECT" compute networks delete "$NAME-network"

The boot disk is retained and continues to incur storage charges. List it with gcloud --project="$PROJECT" compute disks list, snapshot it if needed, and delete it separately only when you no longer need the data. A retained disk is not a database backup.