Google Associate Cloud Engineer Practice Test 10
Question 90
A team needs additional GPU allocation for a short project, but the resource count would exceed its granted quota. What must precede the launch?
- Request additional CPUs only and ignore GPU family quotas.
- Check the GPU-family quota and zonal availability, then request appropriate quota or choose supported capacity.
- Choose a different GPU model without checking that application's accelerator and driver requirements.
- Move the project to a different folder so GPUs no longer count against quotas.
Explanation
Correct Answer
B:GPU launches require both appropriate quota and actual availability in the intended location. Verify the GPU model and quota scope, request an approved adjustment when needed, and select an available supported zone or accelerator if capacity is constrained.
Incorrect Answers
A:GPU allocations can have resource-specific limits separate from CPU allowances.
C:Changing hardware blindly may make the workload incompatible even if quota becomes available.
D:Folder placement cannot itself create physical GPU capacity or remove the project’s enforced quotas.
Question 91
A deployment fails in one zone but quota screens show adequate regional CPU allowance. What other limitation should be inspected?
- Treat regional quota headroom as a guarantee that the requested VM exists in the target zone.
- Increase Cloud Billing budget and retry the same unavailable machine type.
- Change the service account attached to the VM to avoid the zone’s capacity limit.
- Check machine-family availability and zonal capacity or placement restrictions for the requested resource.
Explanation
Correct Answer
D:A request can fail despite sufficient regional quota if the selected zone lacks capacity or does not offer the machine family. Inspect the exact error and supported zones, then adjust placement or capacity planning accordingly.
Incorrect Answers
A:Quota expresses an allowed amount; it is not a reservation of compute stock in every zone.
B:Billing headroom does not provision a machine type that is not available at the target location.
C:Workload identity does not explain a location-specific capacity shortage.
Question 92
An administrator sees a low project quota only for newly enabled API endpoints, not for the related compute service. What should they distinguish?
- Assume API request limits and VM CPU limits are one shared global number.
- Identify whether the exceeded metric is API request rate or compute resource quota, and act on that specific scope.
- Move all workloads to a new billing account to reset service quotas.
- Increase Compute Engine CPU quota when the error identifies a Service Usage API rate limit.
Explanation
Correct Answer
B:A rate limit for an enabled API does not mean the project lacks CPUs. Inspect the actual metric and error, then adjust call patterns or request the appropriate quota instead of changing an unrelated resource limit.
Incorrect Answers
A:API call-rate controls and compute resource quotas measure different activities and may have different scopes.
C:Changing the payer is not a reliable or supported quota-remediation method.
D:The two quotas measure different resources; raising CPU allowance cannot fix the stated API request-rate threshold.
Question 93
A new region cannot host the chosen compute accelerator despite available organization budget. What is the first location check?
- Check whether any accelerator is available somewhere in the region without confirming the required model.
- Request more GPU quota in the target region before verifying the accelerator model is offered there.
- Verify the exact accelerator model’s supported regions and zones before planning the deployment.
- Move the workload to another zone in the same region without checking model support in that zone.
Explanation
Correct Answer
C:Accelerator availability varies by location and machine configuration. Confirm the specific model and region/zone support first, then check quota and capacity for a compatible placement rather than assuming a funded project can deploy anywhere.
Incorrect Answers
A:A region can offer some accelerators but not the exact model or configuration selected for this workload.
B:More quota cannot make a hardware model supported in a location where Google does not offer it.
D:Changing zones blindly can reproduce the same limitation when the requested accelerator is unsupported or unavailable.
Question 94
A deployment requests a quota increase without identifying which service metric is exhausted. Which evidence should be collected first?
- Increase every available quota by the same amount to prevent future errors.
- Raise the project’s alert-only Cloud Billing budget to prove it needs capacity.
- Capture the exact exceeded metric, region or scope, current usage and requested limit.
- Request a global quota increase without reporting the region named in the failed request.
Explanation
Correct Answer
C:A useful quota adjustment request must specify the enforced metric and its scope. Reproduce the failure or inspect the quota page, record current usage and limits, and provide the amount needed before requesting changes.
Incorrect Answers
A:Blanket increases do not target the failing service or resource and may not be approved.
B:A spending budget does not identify which service quota caused the failure.
D:A request that omits the actual enforcement scope may target the wrong quota and does not document the failure.
Question 95
Project atlas has spare CPU quota in europe-west1, but an inherited organization policy permits resource creation only in approved US regions. Which outcome is consistent with both controls?
- Quota headroom causes the location policy to be ignored for a one-time provisioning request.
- The operator can deploy in europe-west1 by granting themselves a higher project-level IAM allow role.
- The deployment must use an approved location or obtain a governed location-policy change; quota alone does not permit europe-west1.
- The policy automatically relocates the requested european VM into us-central1 with identical networking settings.
Explanation
Correct Answer
C:The deployment must use an approved location or obtain a governed location-policy change; quota alone does not permit europe-west1. Resource location constraints are evaluated independently from resource quota. Only an allowed location or authorized policy exception resolves the restriction.
Incorrect Answers
A:Quota headroom causes the location policy to be ignored for a one-time provisioning request. Quotas restrict quantities and have their own scopes; they do not supersede organization policies that forbid resource locations.
B:The operator can deploy in europe-west1 by granting themselves a higher project-level IAM allow role. A project-level allow grant may authorize resource creation but cannot disable a higher-level effective location constraint.
D:The policy automatically relocates the requested european VM into us-central1 with identical networking settings. Organization Policy rejects prohibited configurations rather than transparently moving the VM into a different region and VPC.
Question 96
An operations team can view VM CPU time series in Cloud Monitoring but receives no notification when CPU use stays above its threshold. Which missing monitoring setup provides proactive alerts?
- Create an alerting policy evaluating the appropriate VM CPU utilization metric and attach an approved notification channel.
- Enable a BigQuery billing export; billing SQL queries automatically page VM operators when CPU reaches a threshold.
- Add another read-only dashboard; dashboards create pager notifications automatically without an alert policy.
- Increase the Compute Engine CPU quota; higher quota generates alert incidents and delivers emails to on-call staff.
Explanation
Correct Answer
A:Create an alerting policy evaluating the appropriate VM CPU utilization metric and attach an approved notification channel. An alerting policy evaluates metrics and creates alert incidents when conditions are met; its notification channel determines where a page or email is delivered.
Incorrect Answers
B:Enable a BigQuery billing export; billing SQL queries automatically page VM operators when CPU reaches a threshold. Billing cost exports analyze charges, not metric threshold alert incidents. They do not create CPU incident notifications.
C:Add another read-only dashboard; dashboards create pager notifications automatically without an alert policy. Dashboards visualize collected metrics but do not automatically define thresholds or notification recipients.
D:Increase the Compute Engine CPU quota; higher quota generates alert incidents and delivers emails to on-call staff. CPU quota controls capacity allotment and does not configure a Cloud Monitoring alert or notification destination.
Question 97
A project already emits audit logs, but the incident response team needs approved longer-term storage and a documented routing plan. Which logging configuration should it establish?
- Define the approved log buckets, retention period and routing sinks for the audit evidence.
- Turn off audit logs until the incident response policy is finished.
- Rely only on the current VM image backups to reconstruct all administrative activity.
- Create a Monitoring dashboard and treat charts as an immutable audit log archive.
Explanation
Correct Answer
A:A usable evidence pipeline needs deliberate retention and destinations, not simply events being generated. Configure the supported Cloud Logging buckets and routing controls to meet the organization’s incident retention and access requirements.
Incorrect Answers
B:Disabling audit logging would remove potential evidence rather than preserve it.
C:VM backups cannot replace an audit event history showing authorized administrative operations.
D:Dashboards visualize metrics; they are not an approved long-term store for log entries.
Question 98
A team wants to visualize multiple services' error rates on one operational dashboard without granting each viewer write rights. What should be set up?
- Give every dashboard viewer project Editor so charts can be opened.
- Build a Cloud Monitoring dashboard using the relevant metrics and grant read-only monitoring access.
- Copy all error-rate measurements into the Cloud Billing budget history.
- Configure dashboard charts but leave viewers without the Monitoring permissions to read the underlying time series.
Explanation
Correct Answer
B:A dashboard can present selected service error metrics in one place. Use an appropriate Monitoring viewer role for read-only users and reserve configuration privileges for the operators who maintain the charts.
Incorrect Answers
A:Viewing existing metric charts does not require unrestricted project editing permissions.
C:Billing budgets track costs, not per-service application error time series.
D:A chart's configuration does not give a viewer IAM authorization to read metric data.
Question 99
A new Google Cloud workload is required to use Google Cloud Observability but has no metrics coming from its custom daemon. Which prerequisite should be assessed?
- Create an alert on a nonexistent metric and expect it to generate the daemon’s measurements.
- Instrument or configure the daemon to export supported metrics to Cloud Monitoring before building charts.
- Create a metrics scope across projects without instrumenting the daemon that produces no custom telemetry.
- Grant the project Viewer role to all users, which automatically exports every daemon metric.
Explanation
Correct Answer
B:The custom daemon must expose telemetry through a supported collection mechanism, such as application instrumentation or a metrics ingestion path. Confirm samples appear in Monitoring, then create dashboards and alerts against the actual metric.
Incorrect Answers
A:An alert evaluates incoming time-series data; it does not instrument the application.
C:A metrics scope controls visibility into existing time series, not whether an application emits the missing metric.
D:Viewer controls resource access and is unrelated to collecting the daemon’s missing measurements.
Question 100
A monitoring team needs a cross-project view of related production services under one operational workspace. What configuration should it evaluate?
- Copy every application VM into a single production project to view its metrics.
- Add the production projects to one shared billing budget and expect charts to combine automatically.
- Grant organization Owner to all incident responders without configuring Monitoring.
- Configure an authorized Cloud Monitoring metrics scope that includes the relevant projects.
Explanation
Correct Answer
D:A metrics scope lets a selected scoping project chart and alert on time series from authorized monitored projects. Add the intended production projects and grant users appropriately scoped monitoring roles rather than changing resource ownership.
Incorrect Answers
A:Moving workloads is unnecessary for shared monitoring and can disrupt established boundaries.
B:Cost aggregation does not combine operational time-series data.
C:Broad IAM access does not configure which projects a Monitoring workspace can observe.