Examkin

Soru

Zorluk: Çok zorDeploying and Managing Google Kubernetes Engine (GKE) Clusters and Workloads

An enterprise telemetry engineering team is designing a scalable batch data processing pipeline on Google Cloud. The workload requires custom Linux kernel parameters (`sysctl` settings) on the underlying cluster nodes to handle high network socket churn. The batch processing jobs are fully fault-tolerant and can recover gracefully from node terminations. Additionally, a cloud engineer needs to configure a new administrator workstation to manage workloads on this newly provisioned cluster. Which deployment and configuration strategy satisfies all requirements while following Google Cloud best practices?

  1. A
    Deploy a GKE Autopilot cluster configured with Spot Pods, and execute `kubectl apply -f deployment.yaml` directly from the administrator workstation.
  2. B
    Deploy a GKE Standard cluster with a Spot node pool using custom node configuration flags, set the default cluster property using `gcloud config set container/cluster <cluster-name>`, and execute `kubectl apply -f deployment.yaml`.
  3. Deploy a GKE Standard cluster containing a Spot VM node pool configured with custom `sysctl` node settings, fetch cluster credentials using `gcloud container clusters get-credentials <cluster-name>`, and execute `kubectl apply -f deployment.yaml`.Cevap
  4. D
    Deploy a GKE Standard cluster with a standard, non-preemptible node pool configured with custom sysctl settings, run `kubectl config set-cluster` to establish cluster connectivity, and execute `kubectl apply -f deployment.yaml`.

Cevap

Deploy a GKE Standard cluster containing a Spot VM node pool configured with custom `sysctl` node settings, fetch cluster credentials using `gcloud container clusters get-credentials <cluster-name>`, and execute `kubectl apply -f deployment.yaml`.
The correct strategy leverages GKE Standard to allow custom node kernel modifications (`sysctl`), uses Spot VM node pools to minimize compute cost for fault-tolerant batch jobs, and uses `gcloud container clusters get-credentials` to generate the necessary `kubeconfig` credentials for workstation `kubectl` access.

Adım Adım Çözüm

1
Evaluate operational mode requirements (Autopilot vs. Standard)
GKE Standard must be selected because GKE Autopilot restricts node-level administration, preventing custom Linux kernel `sysctl` parameters.
Autopilot manages the node infrastructure and locks down system kernel modifications for security and reliability.
2
Determine appropriate node pool compute pricing model
Select a Spot VM node pool for the GKE Standard cluster.
Batch processing jobs are fault-tolerant, making them ideal candidates for Spot VMs to reduce compute expenses by up to 60-91%.
3
Configure local administration workstation authentication
Execute `gcloud container clusters get-credentials <cluster-name> --zone <zone>`.
This command fetches cluster API endpoints and OAuth authentication tokens, updating `$HOME/.kube/config` so `kubectl` can authenticate against the control plane.
4
Deploy workload to the cluster
Execute `kubectl apply -f deployment.yaml` successfully.
With valid `kubeconfig` context pointing to a cluster that supports custom node kernel settings, the deployment manifest applies cleanly.

Anahtar Kavram

GKE Cluster Mode Selection, Spot Node Pool Usage, and Kubeconfig Context Generation
Bu soruyu puanla

Konular

Deploying and Managing Google Kubernetes Engine (GKE) Clusters and Workloads