Skip to content

Professional Cloud DevOps Engineer · Domain 1: Establishing and bootstrapping foundational infrastructure and organizational governance · 18% of the exam

Task 1.1: Designing the overall resource hierarchy for a Google Cloud organization.

Folders and projects, shared networking with Shared VPC, peering, Network Connectivity Center and Private Service Connect, multi-project monitoring and logging, IAM and organization policies, service accounts and data residency.

Study it

Sample questions

Try each one before opening the answer. Every option is explained, with the Google Cloud documentation page that proves it.

Question 1 · choose 1

A payments platform runs in 12 projects under one folder, and each project team keeps its logs in its own project's log buckets with its own retention. On-call engineers work from a separate project, ops-oncall, and want the Logs Explorer to open on the logs of all 12 projects at once whenever they open it there, without changing where any log entry is stored. What should the platform team configure?

  1. AAn aggregated sink on the folder that routes the 12 projects' logs into one log bucket in ops-oncall
  2. BThe 12 projects added as monitored projects to the metrics scope of ops-oncall
  3. CA log view on the _Default bucket of each project, granted to the on-call engineers' group
  4. DA log scope in ops-oncall that lists the 12 projects, set as that project's default log scope
Show the answer and why
  • AAn aggregated sink on the folder that routes the 12 projects' logs into one log bucket in ops-oncall

    Incorrect

    A sink changes where entries are stored: it writes them into a central bucket with that bucket's retention, which is centralized storage rather than a read-time view over the teams' own buckets.

  • BThe 12 projects added as monitored projects to the metrics scope of ops-oncall

    Incorrect

    A metrics scope decides which projects' metric data Cloud Monitoring charts and alerts on; it does not decide which resources the Logs Explorer searches.

  • CA log view on the _Default bucket of each project, granted to the on-call engineers' group

    Incorrect

    A log view controls access to a subset of entries in one bucket. On its own it does not make the Logs Explorer in ops-oncall query 12 projects together.

  • DA log scope in ops-oncall that lists the 12 projects, set as that project's default log scope

    Correct

    A log scope lists the resources the Logs Explorer searches and combines the results at read time; the default log scope is what the page searches when it opens.

Log scopes give read-time aggregation of logs stored in different projects: the Logs Explorer queries each listed resource and merges the results, so no entry is copied or moved. Centralized storage with aggregated sinks is the other pattern, and it changes where entries live. Metrics scopes and log views solve different problems (metrics, and access within a bucket).

Question 2 · choose 1

A company is reorganizing around its products. The ordering product's load balancers, GKE Deployments and managed instance groups are spread over four projects, and the operations team wants to define the product as one unit whose services and workloads it can monitor and manage together, with the product's metadata kept in one place. Which resource hierarchy design fits?

  1. AA Shared VPC host project for the four projects, with all of the product's subnets shared from that host
  2. BA product label on every resource in the four projects, with dashboards filtered on that label
  3. CAn app-enabled folder over the four projects, with the product defined as an App Hub application
  4. DA fifth project that holds copies of the product's logs and metrics, routed from the other four
Show the answer and why
  • AA Shared VPC host project for the four projects, with all of the product's subnets shared from that host

    Incorrect

    Shared VPC centralizes network administration; it does not group services and workloads into an application that can be monitored and managed as a whole.

  • BA product label on every resource in the four projects, with dashboards filtered on that label

    Incorrect

    Labels help with filtering and cost reports, but they create no application boundary and no place that stores the application's definition and metadata.

  • CAn app-enabled folder over the four projects, with the product defined as an App Hub application

    Correct

    An app-enabled folder is the application management boundary: its management project stores the application's metadata, and the App Hub application groups the services and workloads.

  • DA fifth project that holds copies of the product's logs and metrics, routed from the other four

    Incorrect

    Copying telemetry into another project gives a shared view of data, not an application made of the product's services and workloads.

The application-centric model groups the services (such as a load balancer's forwarding rule) and workloads (such as a MIG or GKE Deployment) behind one business function into an App Hub application. Making the folder app-enabled gives it a management project that holds the application metadata and lets the team monitor and manage the application at folder level.

Question 3 · choose 1

A network team attached the service project apps-prod to the Shared VPC host project net-host with gcloud commands, not the console. It granted the GKE service agent of apps-prod the Compute Network User role on the subnet that the team will use. Creating a GKE cluster in apps-prod on that subnet still fails because the agent cannot perform network management operations in net-host. Which binding is missing?

  1. ACompute Network User for the same GKE service agent on the whole net-host project instead of on the single subnet
  2. BHost Service Agent User (roles/container.hostServiceAgentUser) on net-host for the GKE service agent of apps-prod
  3. CCompute Network Admin on net-host for the Compute Engine default service account of apps-prod
  4. DHost Service Agent User on apps-prod for the GKE service agent of the net-host project
Show the answer and why
  • ACompute Network User for the same GKE service agent on the whole net-host project instead of on the single subnet

    Incorrect

    Compute Network User may be granted on a specific subnet or on the whole host project; widening its scope does not add the network management permissions the cluster needs.

  • BHost Service Agent User (roles/container.hostServiceAgentUser) on net-host for the GKE service agent of apps-prod

    Correct

    Each service project's GKE service agent needs a binding for roles/container.hostServiceAgentUser on the host project, so that it can perform network management operations there as the host project's GKE service agent would.

  • CCompute Network Admin on net-host for the Compute Engine default service account of apps-prod

    Incorrect

    Network operations for the cluster are done by the GKE service agent of the service project, not by the Compute Engine default service account, and Network Admin is far broader than needed.

  • DHost Service Agent User on apps-prod for the GKE service agent of the net-host project

    Incorrect

    The direction is reversed: the role is granted on the host project to the service project's GKE service agent.

In a Shared VPC, the GKE service agent of each service project needs Compute Network User on the subnet (or the host project) and Host Service Agent User on the host project. The console grants Host Service Agent User automatically when it attaches a service project; with gcloud you grant it yourself.

Question 4 · choose 1

A team's builds and build configuration files already live in me-central2. A sovereignty review now adds that every API request the team's tooling sends to Cloud Build must be processed only in me-central2, with TLS terminated there, and never handled in another location. What should the team change in its gcloud configuration?

  1. AKeep the default global endpoint and add the --region=me-central2 flag to every gcloud builds command the tooling runs
  2. BCreate the build triggers in the global region so that Cloud Build picks the region named in each build configuration file
  3. CEnforce the resource locations organization policy constraint with only me-central2 allowed on the team's project
  4. DSet the gcloud property api_endpoint_overrides/cloudbuild to the endpoint https://cloudbuild.me-central2.rep.googleapis.com/
Show the answer and why
  • AKeep the default global endpoint and add the --region=me-central2 flag to every gcloud builds command the tooling runs

    Incorrect

    Requests sent to a global endpoint can be processed in a location other than the one where the resource resides, so a region flag alone does not give the guarantee.

  • BCreate the build triggers in the global region so that Cloud Build picks the region named in each build configuration file

    Incorrect

    A global trigger only chooses where the build runs; the requests still go through the global endpoint and can be processed elsewhere.

  • CEnforce the resource locations organization policy constraint with only me-central2 allowed on the team's project

    Incorrect

    That constraint limits where new resources can be created; it does not control where API requests are routed and processed.

  • DSet the gcloud property api_endpoint_overrides/cloudbuild to the endpoint https://cloudbuild.me-central2.rep.googleapis.com/

    Correct

    Regional endpoints only let a request proceed if the resource is in the endpoint's location, process it only there and terminate TLS there, which meets data residency and sovereignty requirements.

Cloud Build offers regional endpoints of the form cloudbuild.LOCATION.rep.googleapis.com. Unlike global endpoints, they guarantee that requests are processed only within the endpoint's location. The gcloud CLI uses them when the api_endpoint_overrides/cloudbuild property (or the matching environment variable) points at the regional endpoint.

Question 5 · choose 1

A team stops using the Compute Engine default service account for Cloud Deploy. It creates deploy-dev-sa, sets it as executionConfigs.defaultPool.serviceAccount on the dev target, and grants it Kubernetes Engine Developer and Service Account User in the cluster's project. Render jobs for new releases now fail because the account cannot use the Cloud Storage bucket that holds the pipeline's releases and rendered manifests. Which role should deploy-dev-sa get in the Cloud Deploy project?

  1. ACloud Deploy Runner (roles/clouddeploy.jobRunner)
  2. BCloud Deploy Releaser (roles/clouddeploy.releaser)
  3. CCloud Deploy Service Agent (roles/clouddeploy.serviceAgent)
  4. DCloud Deploy Viewer (roles/clouddeploy.viewer)
Show the answer and why
  • ACloud Deploy Runner (roles/clouddeploy.jobRunner)

    Correct

    Any account set as an execution service account must have Cloud Deploy Runner in the Cloud Deploy project; it includes all the permissions the render account needs, including access to the storage bucket.

  • BCloud Deploy Releaser (roles/clouddeploy.releaser)

    Incorrect

    Releaser is for the identity that calls Cloud Deploy to create releases, not for the account that runs render and deploy jobs.

  • CCloud Deploy Service Agent (roles/clouddeploy.serviceAgent)

    Incorrect

    That role belongs to the Google-managed Cloud Deploy service agent and is not meant for user-managed execution service accounts.

  • DCloud Deploy Viewer (roles/clouddeploy.viewer)

    Incorrect

    Viewer gives read access to Cloud Deploy resources; it lacks the permissions that Cloud Deploy Runner bundles for render jobs.

Cloud Deploy runs render and deploy jobs as the execution service account, which by default is the Compute Engine default service account. When a target names a different account, that account needs Cloud Deploy Runner in the Cloud Deploy project plus runtime permissions, such as Kubernetes Engine Developer and Service Account User for GKE.

Question 6 · choose 1

A platform team keeps one Cloud Deploy delivery pipeline in the project cicd-hub, in us-central1. The dev, staging and production GKE clusters live in the projects app-dev, app-stg and app-prod, in three different regions. Where should the team define the three Cloud Deploy targets?

  1. AIn each application project, in the region of the cluster that the target deploys to
  2. BIn a separate delivery pipeline in each application project, chained to the next one by an automation rule
  3. CIn cicd-hub, in us-central1, each one pointing at its cluster in the other project and region
  4. DIn cicd-hub, with each target created in its cluster's region rather than in us-central1
Show the answer and why
  • AIn each application project, in the region of the cluster that the target deploys to

    Incorrect

    Targets must be defined in the same project and region as the delivery pipeline that uses them, so targets in other projects cannot be stages of the pipeline in cicd-hub.

  • BIn a separate delivery pipeline in each application project, chained to the next one by an automation rule

    Incorrect

    Promotion follows the progression inside one delivery pipeline; splitting the environments into separate pipelines loses that single promotion sequence.

  • CIn cicd-hub, in us-central1, each one pointing at its cluster in the other project and region

    Correct

    Targets live in the pipeline's project and region, while the runtimes they deploy to can be in different projects and regions.

  • DIn cicd-hub, with each target created in its cluster's region rather than in us-central1

    Incorrect

    A target must be in the same region as its delivery pipeline; matching the cluster's region instead breaks that rule.

A Cloud Deploy target is a resource of the delivery pipeline's project and region. The cluster or service that a target deploys to is referenced by its full resource path and may sit in another project and region, which lets a central CI/CD project drive deployments into separate environment projects.

Practice domain 1 →Practice all domains →