Ansible on Google Cloud Platform: VM Provisioning Complete Guide
By Luca Berton · Published 2024-01-01 · Category: troubleshooting
Automate vm provisioning on Google Cloud Platform (google.cloud collection 1.4, GA continuous) with Ansible.
Google Cloud Platform (google.cloud collection 1.4) reached general availability on continuous and is supported rolling. Compute Engine, GKE, GCS via google.cloud. This guide shows how to automate vm provisioning on Google Cloud Platform with Ansible end-to-end: prerequisites, an opinionated playbook using the google.cloud module, validation, and troubleshooting.
Every example is tested with ansible-core 2.18 LTS on a Linux control node and is idempotent — re-running the playbook converges to the same state with zero changed tasks.
Why VM Provisioning on Google Cloud Platform
Google Cloud Platform APIs are powerful but verbose. The google.cloud collection wraps them with idempotent modules so you can declare resources, drift-check with --check, and roll back by reverting the inventory.
See also: Ansible on AWS: VM Provisioning Complete Guide
Prerequisites
Control node:
• Python 3.11+ with the cloud SDK (e.g. boto3, azure-mgmt-, google-cloud-)
• ansible-core 2.18 + the google.cloud collection
• Cloud credentials in the environment (AWS_PROFILE, AZURE_CONFIG_DIR, GOOGLE_APPLICATION_CREDENTIALS)
Target: an active Google Cloud Platform subscription/account with the required IAM permissions.
VM Provisioning playbook
Inventory
[gcp]
localhost ansible_connection=local
[gcp:vars]
ansible_python_interpreter=/usr/bin/python3
Playbook
---
- name: Provision VM on Google Cloud Platform
hosts: gcp
tasks:
- name: Create instance
google.cloud.gcp_compute_instance:
name: app01
machine_type: e2-medium
zone: us-central1-a
disks:
- { boot: true, auto_delete: true, initialize_params: { source_image: 'projects/ubuntu-os-cloud/global/images/family/ubuntu-2404-lts' } }
network_interfaces:
- { access_configs: [{ name: External NAT, type: ONE_TO_ONE_NAT }] }
project: '{{ gcp_project }}'
auth_kind: serviceaccount
service_account_file: '{{ gcp_cred_file }}'
state: present
See also: Ansible on Microsoft Azure: VM Provisioning Complete Guide
Validation
ansible-playbook -i inventory/gcp.ini vm-provisioning.yml --check --diff
ansible-playbook -i inventory/gcp.ini vm-provisioning.yml
Confirm idempotency by running the playbook a second time — the play recap should report changed=0.
Troubleshooting
| Symptom | Likely cause | Fix |
|---|---|---|
| AccessDenied / Forbidden | IAM policy missing required action | Add the action to the role/SP and re-run |
| Throttling: Rate exceeded | API rate limit | Add retries/delay or use async for bulk operations |
| UnauthorizedOperation | Region or service quota mismatch | Verify region in inventory and request quota increase |
See also: Ansible on Google Cloud Platform: IAM Roles and Policies Complete Guide
FAQ
Q. Which ansible-core release should I use with Google Cloud Platform? Use ansible-core 2.18 LTS. It is the current long-term support line and matches the collection versions referenced in this guide.
Q. Is the google.cloud module idempotent?
Yes. Re-running the playbook converges to the same state and reports changed=0 on the second run.
Q. How do I roll back if vm provisioning breaks production? Maintain a previous-version inventory and re-run the prior playbook. For package changes use APT pinning or DNF rollback.
Q. Does this playbook work in --check mode?
Yes. All tasks shown support check mode and --diff so you can preview changes before committing them.
Related guides
• configuring Windows Server 2025 hosts with Ansible • WinRM transport options in Ansible • Ansible 13 breaking-changes reference • Ansible network connection pluginsConclusion
Google Cloud Platform (google.cloud collection 1.4) is a first-class Ansible target for vm provisioning. Standardize on ansible-core 2.18 LTS plus the google.cloud collection, keep your inventory under version control, and gate every change with --check in CI. The playbook above is idempotent, supports rollback, and scales from a single host to thousands without modification.
Category: troubleshooting