gcloud compute resource-policies create group-placement

NAME: gcloud compute resource-policies create group-placement - create a Compute Engine group placement resource policy
SYNOPSIS: gcloud compute resource-policies create group-placement NAME [ --accelerator-topology-mode = ACCELERATOR_TOPOLOGY_MODE ] [ --availability-domain-count = AVAILABILITY_DOMAIN_COUNT ] [ --collocation = COLLOCATION ] [ --description = DESCRIPTION ] [ --gpu-topology = GPU_TOPOLOGY ] [ --region = REGION ] [ --vm-count = VM_COUNT ] [ GCLOUD_WIDE_FLAG … ]
DESCRIPTION: Create a Compute Engine Group Placement Resource Policy.
EXAMPLES: To create a Compute Engine group placement policy with two availability domains, run:
gcloud compute resource-policies create group-placement my-resource-policy --region = REGION --availability-domain-count = 2
POSITIONAL ARGUMENTS: NAME

Name of the resource policy to operate on.
FLAGS: --accelerator-topology-mode = ACCELERATOR_TOPOLOGY_MODE

Defines the accelerator connection strategy for accelerator machine types like TPUs. ACCELERATOR_TOPOLOGY_MODE must be one of:

AUTO_CONNECT

This creates a static, pre-formed accelerator topology.

PROVISION_ONLY

The interconnected chips are connected on demand. At the time of VM creation, the chips are not connected.

--availability-domain-count = AVAILABILITY_DOMAIN_COUNT

Number of availability domain in the group placement policy.

--collocation = COLLOCATION

Collocation specifies whether to place VMs inside the sameavailability domain on the same low-latency network. COLLOCATION must be one of:

collocated

Low network latency between more VMs placed on the same availability domain.

unspecified-collocation

Unspecified network latency between VMs placed on the same availability domain. This is the default behavior.

--description = DESCRIPTION

An optional, textual description for the backend.

--gpu-topology = GPU_TOPOLOGY

Specifies the shape of the GPU slice, in slice based GPU families eg. A4X.

--region = REGION

Region of the resource policy to operate on. If not specified, you might be prompted to select a region (interactive mode only).
To avoid prompting when this flag is omitted, you can set the compute/region property:

gcloud config set compute/region REGION

A list of regions can be fetched by running:

gcloud compute regions list

To unset the property, run:

gcloud config unset compute/region

Alternatively, the region can be stored in the environment variable CLOUDSDK_COMPUTE_REGION .

--vm-count = VM_COUNT

Number of instances targeted by the group placement policy. Google does not recommend that you use this flag unless you use a compact policy and you want your policy to work only if it contains this exact number of VMs.
GCLOUD WIDE FLAGS: These flags are available to all commands: --access-token-file , --account , --billing-project , --configuration , --flags-file , --flatten , --format , --help , --impersonate-service-account , --log-http , --project , --quiet , --trace-token , --user-output-enabled , --verbosity .
Run $ gcloud help for details.
NOTES: These variants are also available:
gcloud alpha compute resource-policies create group-placement

gcloud beta compute resource-policies create group-placement

gcloud preview compute resource-policies create group-placement

gcloud compute resource-policies create group-placement Stay organized with collections Save and categorize content based on your preferences.

gcloud compute resource-policies create group-placement