An Associate Cloud Engineer is managing a specialized analytics application deployed on a Compute Engine virtual machine in the us-central1-a zone. To process a new data pipeline, the engineer updates the VM configuration to attach an NVIDIA GPU. When attempting to start the VM, the operation fails with a QUOTA_EXCEEDED error for GPU resources in that zone. The application must remain in us-central1-a due to strict low-latency requirements with adjacent infrastructure. Which action should the engineer take to resolve this issue following Google-recommended practices?
- Submit a quota increase request for the required GPU resource type in the us-central1 region via the Google Cloud Console.Cevap
- BDeploy an additional VM in another region and peer the VPC networks to bypass the local zonal GPU limit without requesting additional allocation.
- CChange the VM provisioning model to a Spot VM instance to automatically bypass GPU quota restrictions.
- DGrant the primitive Owner role to the Compute Engine default service account so that instance startup operations bypass project quota checks.
Cevap
Submit a quota increase request for the required GPU resource type in the us-central1 region via the Google Cloud Console.
When a Compute Engine instance fails to launch due to a QUOTA_EXCEEDED error, the official Google Cloud practice is to request a quota limit increase via the Cloud Console Quotas page or gcloud CLI. This allows Google Cloud to evaluate the request and raise the limit for the target region.
Adım Adım Çözüm
Anahtar Kavram
Managing Compute Engine Quotas and Zonal Resource Allocations