Hands-on Lab — ACE Google Cloud Associate Cloud Engineer

Overview

Prerequisites

Steps

1. 1.Provider, project services, naming

   Enable Compute Engine, IAM, and Cloud Logging APIs. ACE exam tests this enable-API-per-resource pattern repeatedly.

   Copy

   terraform {
     required_version = ">= 1.5"
   
     required_providers {
       google = { source = "hashicorp/google", version = "~> 6.0" }
     }
   }
   
   provider "google" {
     project = "your-project-id" # REPLACE
     region  = "us-central1"
     zone    = "us-central1-a"
   }
   
   locals {
     labels = {
       project    = "certlabpro-ace"
       managed_by = "terraform"
     }
   }
   
   resource "google_project_service" "compute" {
     service            = "compute.googleapis.com"
     disable_on_destroy = false
   }
   
   resource "google_project_service" "iam" {
     service            = "iam.googleapis.com"
     disable_on_destroy = false
   }
   
   resource "google_project_service" "logging" {
     service            = "logging.googleapis.com"
     disable_on_destroy = false
   }
   
   resource "google_project_service" "monitoring" {
     service            = "monitoring.googleapis.com"
     disable_on_destroy = false
   }

2. 2.Carve a custom VPC + regional subnet

   Provisions:

   • VPC
   • Compute Engine

   GCP networks are global — a single VPC spans all regions, and subnets are regional primitives carved out of it. This is a recurring ACE exam-question shape: AWS has regional VPCs; Azure has regional VNets; GCP has a global VPC with regional subnets.

   We disable auto_create_subnetworks (custom-mode VPC, the ACE-recommended default) and carve a single /24 subnet in us-central1 with Private Google Access enabled so VMs without external IPs can still reach GCP APIs.

   Copy

   resource "google_compute_network" "main" {
     name                    = "certlabpro-ace-vpc"
     auto_create_subnetworks = false
     routing_mode            = "REGIONAL"
   
     depends_on = [google_project_service.compute]
   }
   
   resource "google_compute_subnetwork" "main" {
     name                     = "certlabpro-ace-subnet"
     ip_cidr_range            = "10.10.1.0/24"
     region                   = "us-central1"
     network                  = google_compute_network.main.id
     private_ip_google_access = true
   }

3. 3.Add two firewall rules (internal + IAP SSH)

   Provisions:

   • VPC

   GCP firewall rules are VPC-scoped (not per-subnet) and have explicit direction: ingress (default) or egress. Default ingress is deny-all; default egress is allow-all. ACE exam tests this direction + default-deny invariant relentlessly.

   We add:

   1. Allow internal — TCP/UDP/ICMP between VMs in the 10.10.1.0/24 subnet.
   2. Allow IAP SSH — TCP/22 from Google's IAP gateway range (35.235.240.0/20) so you can SSH without exposing the VM to the internet.

   The IAP-SSH rule is the ACE-recommended pattern — it requires roles/iap.tunnelResourceAccessor on the VM (binding handled in Step 4).

   Copy

   resource "google_compute_firewall" "allow_internal" {
     name      = "certlabpro-ace-allow-internal"
     network   = google_compute_network.main.name
     direction = "INGRESS"
   
     source_ranges = ["10.10.1.0/24"]
   
     allow {
       protocol = "tcp"
     }
     allow {
       protocol = "udp"
     }
     allow {
       protocol = "icmp"
     }
   }
   
   resource "google_compute_firewall" "allow_iap_ssh" {
     name      = "certlabpro-ace-allow-iap-ssh"
     network   = google_compute_network.main.name
     direction = "INGRESS"
   
     source_ranges = ["35.235.240.0/20"] # IAP gateway range
   
     allow {
       protocol = "tcp"
       ports    = ["22"]
     }
   }

4. 4.Provision a GCE VM with a least-privilege service account

   Provisions:

   • Compute Engine
   • IAM

   Compute Engine VMs run with an attached service account — the GCP equivalent of an AWS EC2 instance profile or an Azure VM managed identity. ACE-recommended pattern: never use the default Compute service account (over-privileged); always create a per-workload service account with only the IAM roles the workload needs.

   We create a service account, grant it roles/logging.logWriter + roles/monitoring.metricWriter (the minimum for the Ops Agent to send logs / metrics), and attach it to an e2-micro Debian VM with no external IP. The IAP-SSH firewall rule from Step 3 lets you reach it via tunnel.

   Copy

   resource "google_service_account" "vm" {
     account_id   = "certlabpro-ace-vm-sa"
     display_name = "ACE lab VM service account"
   }
   
   resource "google_project_iam_member" "vm_log_writer" {
     project = data.google_project.current.project_id
     role    = "roles/logging.logWriter"
     member  = "serviceAccount:${google_service_account.vm.email}"
   }
   
   resource "google_project_iam_member" "vm_metric_writer" {
     project = data.google_project.current.project_id
     role    = "roles/monitoring.metricWriter"
     member  = "serviceAccount:${google_service_account.vm.email}"
   }
   
   data "google_project" "current" {}
   
   resource "google_compute_instance" "vm" {
     name         = "certlabpro-ace-vm"
     machine_type = "e2-micro"
     zone         = "us-central1-a"
   
     boot_disk {
       initialize_params {
         image = "debian-cloud/debian-12"
       }
     }
   
     network_interface {
       subnetwork = google_compute_subnetwork.main.id
       # No access_config block = no external IP. SSH via IAP only.
     }
   
     service_account {
       email  = google_service_account.vm.email
       scopes = ["cloud-platform"] # API scope; IAM does the real gating
     }
   
     labels = local.labels
   }

5. 5.Wire a Cloud Logging metric + alert policy

   Provisions:

   • Cloud Logging
   • Cloud Monitoring

   ACE-recommended observability shape: a log-based metric counts matches of a Cloud Logging filter, then a Cloud Monitoring alert policy fires when the metric crosses a threshold. We watch for severity >= ERROR log lines from any GCE VM in this project, count them per minute, and alert when more than 5 hit in a 5-minute window.

   The notification channel is left implicit — alert policies need a notification_channels array, but the channel itself (email, Slack, PagerDuty) is typically configured once per project in the console. ACE exam tests this log-metric → alert policy → notification channel triangle as the standard observability primitive.

   Copy

   resource "google_logging_metric" "vm_errors" {
     name   = "certlabpro_ace_vm_errors"
     filter = "resource.type=\"gce_instance\" AND severity >= ERROR"
   
     metric_descriptor {
       metric_kind = "DELTA"
       value_type  = "INT64"
     }
   
     depends_on = [google_project_service.logging]
   }
   
   resource "google_monitoring_alert_policy" "vm_error_burst" {
     display_name = "ACE lab — VM error burst"
     combiner     = "OR"
   
     conditions {
       display_name = "More than 5 ERROR lines per 5 minutes"
   
       condition_threshold {
         filter          = "metric.type=\"logging.googleapis.com/user/${google_logging_metric.vm_errors.name}\" AND resource.type=\"gce_instance\""
         duration        = "300s"
         comparison      = "COMPARISON_GT"
         threshold_value = 5
   
         aggregations {
           alignment_period   = "60s"
           per_series_aligner = "ALIGN_DELTA"
         }
       }
     }
   
     # notification_channels = [] # add channels via the console or as a separate TF resource
   
     depends_on = [google_project_service.monitoring]
   }

Cleanup