Google Associate Cloud Engineer Practice Test 17
Question 160
A CI workflow enables a newly adopted Google API but fails only in the production project due to an organization restriction. What should the release check?
- Inspect the applicable organization service restrictions and approved exception path before retrying enablement.
- Grant project Editor repeatedly until the policy restriction disappears.
- Move the pipeline’s logs to a different bucket, then retry without a governance review.
- Disable the production billing account because billing controls all API allowlists.
Explanation
Correct Answer
A:A permission to call Service Usage does not override effective organization policies. Identify the exact constraint or denied service, obtain any authorized policy exception and reattempt through the controlled production workflow.
Incorrect Answers
B:Broader project IAM cannot negate a prohibition imposed by a higher-level organization policy.
C:Log destination is unrelated to service-use organization constraints.
D:Removing billing can break workloads and does not solve a governance-level API restriction.
Question 161
A company must run an unchanged legacy OS service that needs kernel drivers and persistent VM-level access. Which Google Cloud compute platform preserves that control?
- Deploy it to Cloud Run and expect root-level kernel module installation inside the managed runtime.
- Run it on a suitably configured Compute Engine VM with the required OS and driver control.
- Use a Cloud Storage bucket as the host for the executing operating-system service.
- Use BigQuery scheduled queries as a replacement for a persistent kernel-level daemon.
Explanation
Correct Answer
B:Compute Engine provides VM guest OS administration and supports selecting compatible machine images and drivers. A persistent service requiring kernel-level access can run there with approved patching, network and security controls.
Incorrect Answers
A:Cloud Run abstracts host OS control and is unsuitable for installing custom kernel drivers.
C:Object storage holds data but does not provide a guest OS process host.
D:SQL scheduling is not a host operating-system environment for legacy drivers.
Question 162
A stateless HTTP endpoint arrives in bursts and its owners do not want to manage Kubernetes nodes or VM operating systems. Which deployment model fits?
- Deploy the HTTP container to Cloud Run and configure scaling/concurrency for bursty traffic.
- Provision a permanent Compute Engine fleet and manually scale each instance for requests.
- Create a GKE Standard cluster and manage worker nodes despite the requirement to avoid node administration.
- Store the HTTP handler code in Cloud Storage and expect objects to execute requests directly.
Explanation
Correct Answer
A:Cloud Run supports request-driven container execution without managing VM operating systems or Kubernetes nodes. For a stateless endpoint with variable traffic, configure supported scaling and concurrency according to application behavior.
Incorrect Answers
B:VMs can serve HTTP but introduce the infrastructure operations the team wants to avoid.
C:GKE adds cluster operations for an application that does not need Kubernetes control-plane features.
D:Cloud Storage objects do not execute an HTTP request handler by themselves.
Question 163
A platform team needs fine-grained Kubernetes Deployments, custom controllers, and services across container workloads. Which service should it choose?
- Use Google Kubernetes Engine and manage Deployments, custom resources and controllers with Kubernetes APIs.
- Deploy every workload as a Cloud Billing budget because budgets control autoscaling.
- Use only Cloud Run HTTP services and assume they expose arbitrary Kubernetes controller APIs.
- Create Compute Engine instances and assume Kubernetes Deployments exist without a cluster.
Explanation
Correct Answer
A:GKE provides managed Kubernetes capabilities, including declarative workloads and extensible controllers. It matches a platform that explicitly requires Kubernetes objects and custom orchestration rather than just running a container endpoint.
Incorrect Answers
B:A budget is a financial alert construct, not a workload orchestrator.
C:Cloud Run abstracts Kubernetes cluster administration and is not a general-purpose custom-controller host.
D:VMs can host software, but Kubernetes resource management requires a configured cluster.
Question 164
A nightly containerized reconciliation task must start on a schedule, run until its work finishes, then stop. It does not serve HTTP requests between runs. Which Cloud Run execution type fits?
- Create a Cloud Run job and invoke it on the approved schedule using a supported job scheduler or workflow.
- Deploy a Cloud Run service that must keep an inbound HTTP listener active and continuously hold one warm instance for a one-off batch.
- Create a Cloud Run service revision without a service handler; the service automatically runs and exits on the schedule.
- Create an external HTTPS load balancer; its frontend is the managed scheduler that executes batch containers.
Explanation
Correct Answer
A:Create a Cloud Run job and invoke it on the approved schedule using a supported job scheduler or workflow. Cloud Run jobs execute containers as tasks until they finish and do not need an HTTP serving handler. They support scheduled executions.
Incorrect Answers
B:Deploy a Cloud Run service that must keep an inbound HTTP listener active and continuously hold one warm instance for a one-off batch. A request-serving service is a different runtime model; keeping an idle listener warm does not match the required finish-and-exit batch execution.
C:Create a Cloud Run service revision without a service handler; the service automatically runs and exits on the schedule. A Cloud Run service expects a serving workload and revisions do not inherently define a scheduled batch job.
D:Create an external HTTPS load balancer; its frontend is the managed scheduler that executes batch containers. A load balancer routes client requests to serving backends and is not a container job scheduling and completion service.
Question 165
A user action emits a Pub/Sub message and a stateless handler must run only when matching messages arrive. Which integration fits the event pattern?
- Poll the Pub/Sub topic from a permanently running VM, regardless of whether events arrive.
- Store each Pub/Sub message in Cloud Storage and wait for the bucket to execute it.
- Create a regional CPU budget so messages automatically invoke the handler.
- Configure a supported Pub/Sub/Eventarc event trigger for a Cloud Run handler with authorized delivery identity.
Explanation
Correct Answer
D:An event trigger decouples the producer from execution and invokes Cloud Run when matching messages are received. Configure event filters, permissions and retries appropriately for at-least-once delivery semantics.
Incorrect Answers
A:A continuously running poller can work but does not meet the stated goal of only running the stateless handler when events arrive.
B:Object storage does not execute messages as application code on its own.
C:Resource budgets do not connect Pub/Sub delivery to compute execution.
Question 166
A VM is created in the default zone but must reside near an existing zonal disk in another zone. What location constraint must the operator respect?
- Create the VM in the disk’s zone, or choose a supported storage migration or replication design.
- Keep the VM in its default zone; a disk name makes it attachable from any zone.
- Turn on a global IP address so the VM can attach the disk across zones.
- Change only the disk’s display label to the target VM zone.
Explanation
Correct Answer
A:Zonal Persistent Disk attachment is constrained by the zone where the disk exists. Create a compatible VM there, or move/replicate data using an explicitly supported approach. Proximity by itself is not the rule; disk location and attachment eligibility are.
Incorrect Answers
B:An instance in another zone cannot ordinarily attach the existing zonal disk simply by referencing it by name.
C:IP addressing affects networking and does not change a block disk’s physical zone or attachment compatibility.
D:Labels are descriptive metadata. They do not relocate a disk or make cross-zone attachment supported.
Question 167
A deployment requests a particular Compute Engine machine series in a zone where Google does not offer it. General CPU quota remains unused. What change addresses the resource availability failure?
- Increase the organization's Cloud Billing budget; payment headroom makes unavailable machine families appear in every zone.
- Create a project IAM Editor binding; broad roles change the geographic product catalog even for unsupported machine types.
- Select a supported machine type in the chosen zone or move to an approved zone where that series is offered.
- Rename the VM to the machine series name; matching names override Compute Engine regional availability checks.
Explanation
Correct Answer
C:Select a supported machine type in the chosen zone or move to an approved zone where that series is offered. Machine series availability is location-dependent. Choosing a supported zone or compatible machine type resolves the actual launch restriction.
Incorrect Answers
A:Increase the organization's Cloud Billing budget; payment headroom makes unavailable machine families appear in every zone. Budget availability does not change the machine type support matrix for a geographic location.
B:Create a project IAM Editor binding; broad roles change the geographic product catalog even for unsupported machine types. Resource permission cannot cause a machine family to become offered where Google Cloud does not support it.
D:Rename the VM to the machine series name; matching names override Compute Engine regional availability checks. Instance labels and names are not used to bypass the product catalog's machine-type/zone availability validation.
Question 168
A production VM must maintain an external address even after stop/start events, and the application currently uses an ephemeral external IP. What should change?
- Add a new ephemeral IPv6 address and assume it stays assigned indefinitely.
- Use an internal DNS name as a substitute for the public endpoint requested by clients.
- Set the VM to restart automatically and keep the current ephemeral external address.
- Reserve a static external IP of the required type and attach it to the instance.
Explanation
Correct Answer
D:A reserved static external address is intended for a stable public endpoint. Reserve the correct regional or global address type for the resource, attach it, and manage its allocation and cost explicitly.
Incorrect Answers
A:An additional ephemeral address still lacks a deliberate reservation for a stable endpoint.
B:Private name resolution does not preserve the public IP address expected by existing external clients.
C:Automatic restart addresses VM recovery, not guaranteed retention of an ephemeral public address through all lifecycle events.
Question 169
A startup script fails on a new VM because its deployment image lacks one required OS package, although the VM boots. What artifact should be corrected?
- Increase VM vCPUs without examining the guest package installation logs.
- Change the VPC firewall rule because the guest operating system installed without errors.
- Fix the startup script or tested base image so the application dependency is installed before launch.
- Disable the startup script, because the VM booted successfully without the package.
Explanation
Correct Answer
C:Inspect the startup-script output and package availability, then correct the installation sequence or bake the dependency into a validated image. Re-test a fresh VM to confirm the application, not just the OS, starts successfully.
Incorrect Answers
A:More processor capacity may make the VM faster, but it cannot install a missing guest package or repair a broken startup script.
B:A firewall may affect package download reachability, but a booted VM with a missing package first requires checking guest install steps and dependencies.
D:Removing the script avoids the failing step but leaves the application missing its required package.
Question 170
A VM needs to survive occasional host maintenance events with minimal interruption but is configured to terminate on host maintenance. What policy should be reviewed?
- Leave the VM set to TERMINATE; the platform will migrate it automatically anyway.
- Increase the network bandwidth to prevent infrastructure maintenance.
- Review the instance’s on-host-maintenance and automatic-restart settings and choose supported behavior for its machine type.
- Move the VM’s billing account whenever Google announces maintenance.
Explanation
Correct Answer
C:Where the machine configuration supports live migration, set on-host-maintenance behavior accordingly and verify the separate automatic-restart requirement. Some VM types cannot live-migrate, so recovery planning must reflect actual support.
Incorrect Answers
A:A terminate-on-maintenance setting does not by itself request the lower-interruption migration behavior.
B:Network throughput cannot prevent planned infrastructure maintenance or alter the host scheduling policy.
D:Billing account association has no effect on how a VM responds to host maintenance.