Prepare and test your skills

Prepare and test your skills

Namespace and Scope Identification is the practice of knowing which level an object belongs to before counting it in inventory. In Google Kubernetes Engine (GKE), the widest scope is the fleet (opens in a new tab), a logical grouping of Kubernetes clusters and other resources that can be managed together. Below the fleet are individual clusters in Google Cloud projects, and within a fleet are team partitions that slice visibility by application team. Identifying the correct scope prevents missing objects that live in a different project, cluster, or team partition.
Fleet scope is the top inventory boundary for multi-cluster operations. The fleet is managed by the Fleet service, also known as the Hub service, and it is viewed from a single fleet host project. For example, ten Google Cloud projects with two GKE clusters each can be grouped into one fleet, so an operator can observe and change twenty clusters from one place instead of switching context between projects. Fleet scope abstracts away cluster boundaries when fleet features are enabled, and a fleet can include only GKE clusters on Google Cloud or also include clusters outside Google Cloud.
Cluster scope is the level below fleet scope, and a cluster only appears in fleet inventory after registration. GKE clusters on Google Cloud must be explicitly registered, either during creation or afterward by quick registration in the console, the gcloud CLI, Terraform, or Config Connector. Clusters on Google Cloud can be registered to a fleet in their current project or, with extra permissions, to a fleet in a different project. Clusters outside Google Cloud join the fleet differently and depend on the Connect Agent for connectivity.
The Connect Agent is installed on the outside cluster to handle communication between the cluster and the Google Cloud project. Traffic flows from the Connect Agent on the cluster to Google Cloud, and the agent can traverse NATs, egress proxies, VPNs, and other interconnects. Clusters using the Connect Agent do not need public or externally exposed IP addresses for their Kubernetes API servers.
The table below compares how different cluster types join the fleet and what registration limits apply to each type.
| Cluster type | How the cluster joins the fleet | Registration constraint |
|---|---|---|
| GKE on Google Cloud | Explicit registration during or after creation | Can register in same project or different project with extra permissions |
| Google Distributed Cloud on-premises | Automatic registration at creation through the cluster configuration file | Cannot be unregistered, except unregistered admin clusters can be added by update |
| GKE on AWS and Azure | Automatic registration at creation through the GKE Multi-Cloud API | Cannot be unregistered or manually registered |
| Third-party attached clusters | Manual attachment through the GKE Multi-Cloud API for conformant clusters with x86 nodes, including EKS and AKS | Incurs a per-vCPU charge under GKE pricing |
Membership location adds another scope detail that controls where fleet inventory is managed. Most fleet members are managed by a global Fleet service, while some GKE clusters on Google Cloud can use regional membership management for situations such as data that cannot leave a region. GKE clusters registered by using Config Connector can use either global or regional management, and all clusters registered before January 2023 use global management. A cluster's membership location cannot be changed while it remains a member, so the cluster must first be unregistered and then registered again to change it.
GKE clusters on Google Cloud also support lightweight memberships, which register the cluster without enabling full fleet configuration. A cluster with lightweight membership can still use fleet-based rollout sequencing for cluster upgrades. Lightweight memberships prevent the cluster from using:
Exam tip: GKE clusters on Google Cloud require explicit registration, while Google Distributed Cloud and GKE on AWS and Azure register automatically at creation and cannot be unregistered.
Team scopes are enterprise fleet-level partitions that associate subsets of fleet member clusters with specific application teams. A single fleet member cluster can be associated with one or more team scopes within its fleet. Team scopes sit between fleet-wide visibility and single-cluster inventory, so operators can locate objects by team instead of checking every cluster. Team scopes enable team-based features across the associated clusters, including access control, team-scoped observability, and sequencing cluster upgrade rollouts. Because the scope is defined once at the fleet level, the same team partition applies consistently across all clusters linked to that team.
Top-down hierarchy from fleet host project to fleet to Team Scope A and B, each linked to its member GKE clusters for team-partitioned inventory.
Command-Line Resource Inspection is the practice of using kubectl (opens in a new tab) to query the live inventory of a Google Kubernetes Engine (GKE) cluster. The kubectl tool sends requests to the cluster control plane and returns the current state, configuration, and events for resources such as nodes, Pods, and Services. This direct view helps explain why a problem is happening, not just that something looks unhealthy. Traffic flows from kubectl on the workstation to the GKE control plane, and the control plane returns only the resources the user is authorized to see.
Connecting kubectl to the cluster is the setup process of installing the tool, pointing it at the right control plane, and confirming permissions. Until this setup is complete, kubectl cannot retrieve node, Pod, or Service data from GKE. The connection depends on Google Cloud credentials and Kubernetes role-based permissions.
gcloud container clusters get-credentials CLUSTER_NAME --location=LOCATION so kubectl communicates with the correct cluster control plane in the correct region or zone.kubectl auth can-i get nodes or a similar check for Pods and Services. A response of yes means the query is allowed, while no or a Forbidden error means a cluster administrator must grant the needed role.The kubectl get command provides an overview of what is running in the cluster. The kubectl get command lists resources in a table that includes names, status, age, and version information. Use this command first to check overall system integrity and resource availability before asking for details about one object. The same get pattern applies to nodes, Pods, and Services when checking the current inventory.
| Resource | Inspection command | Healthy state |
|---|---|---|
| Nodes | kubectl get nodes | Ready |
| Pods | kubectl get pods --all-namespaces | Running |
Any node status other than Ready requires additional investigation, since the node is not available for normal work. Any Pod status other than Running also requires additional investigation, such as a Pod in Pending that is still waiting to be scheduled onto a node.
Exam tip: A node is healthy only when its STATUS is Ready, while a Pod is healthy only when its STATUS is Running — do not treat the two status values as interchangeable.
The kubectl describe function examines the detailed configuration and lifecycle information for one specific node, Pod, or Service. While get shows the overall inventory list, describe shows the selected resource's metadata, current configuration, status conditions, and recent events. This deeper view connects the resource to its dependencies, such as which node hosts a Pod or which configuration is currently applied. Use describe after get shows a resource that is unhealthy or unexpected.
Cloud Console Inventory Management is the visual way to check what is running in Google Kubernetes Engine (GKE) without typing commands. The Google Cloud console provides graphical list and detail views for clusters, while the gcloud CLI provides the same control from the command line. For day-to-day operation, the console is used to see resource distribution across clusters and to open a specific cluster to review its nodes and workloads. This flow moves from fleet-level distribution to cluster-level detail: first find the cluster, then open it, then inspect its configuration.
GKE cluster building blocks are the control plane, the nodes, and the workloads that run on those nodes. A Google Kubernetes Engine (GKE) cluster consists of at least one cluster control plane machine and multiple worker machines called nodes. Nodes are Compute Engine virtual machine (VM) instances that run the Kubernetes processes necessary to make them part of the cluster. You deploy applications to clusters, and the applications run on the nodes, using Pod IP address ranges for Pod-to-Pod communication within the cluster.
How the console represents attached clusters is based on an AttachedCluster resource and a secure tunnel from the external cluster to Google Cloud. GKE attached clusters (opens in a new tab) lets you register any conformant Kubernetes cluster, with support for Amazon EKS clusters and Azure AKS clusters, so it appears in the Google Cloud resource hierarchy. When you install GKE attached clusters, a new attached cluster resource is created within your Google project as a regional resource, and all metadata associated with the cluster is stored in the Google Cloud region associated with that resource, which permits compliance with data residency requirements.
GKE attached clusters deploys a workload called the Connect Agent in your external cluster. The Connect Agent connects back to the Connect API service on Google Cloud to create a secure link to Google Cloud. GKE attached clusters uses this secure tunnel to manage additional features for your clusters, so the console can control and monitor a cluster even when the cluster runs outside Google Cloud.
Central operation through that model enables fleet-wide tools from one interface:
Because the AttachedCluster resource, the Connect Agent, and the secure tunnel work together, the console can show external clusters alongside native GKE clusters for inventory and control.
Opening cluster inventory in the console is the navigation path from the GKE dashboard list view to a specific cluster detail view. The Google Kubernetes Engine dashboard in the console is the starting point for visually monitoring workloads and Services. The dashboard presents a cluster list, and selecting a cluster opens its detail view where sections such as Networking can be reviewed and edited. This same navigation pattern is used for inventory checks and for configuration, so learning to move from list to detail view is the foundation for monitoring nodes, Pods, and Services.
To open an existing cluster in the console:
A node is healthy only when its status is Ready, while a Pod is healthy only when its status is Running. Any other status requires additional investigation, since a node that is not Ready is not available for normal work and a Pod that is not Running may still be waiting to be scheduled onto a node, such as a Pod in Pending.
Google Kubernetes Engine clusters on Google Cloud must be explicitly registered to a fleet during or after creation in the same project or in a different project with extra permissions. In contrast, Google Distributed Cloud on-premises clusters and Google Kubernetes Engine on AWS and Azure clusters register automatically at creation and cannot be unregistered, except unregistered admin clusters for Google Distributed Cloud can be added by update.
A fleet is a logical grouping of Kubernetes clusters and other resources that can be managed together and is the top inventory boundary for multi-cluster operations. The fleet is managed by the Fleet service, also known as the Hub service, and is viewed from a single fleet host project, so multiple clusters across multiple Google Cloud projects can be observed and changed from one place instead of switching context between projects.
Your team recently deployed a containerized microservice application to a Google Kubernetes Engine (GKE) cluster. You need to inspect the operational health of the application, verify that all Pod replicas are running as expected, and investigate any container restart errors or failed deployments using the graphical interface.
Which section under Kubernetes Engine in the Google Cloud Console should you navigate to view this inventory and health status?
Last updated