Create sole-tenant node templates

Console

1. In the Google Cloud console, go to the Sole-tenant nodes page.

   Go to Sole-tenant nodes

2. Click Create node group. The process of creating a node group in the console includes creating or selecting a node template.

3. Specify a Name for the node group.

4. Specify a Region to create the node template in. You can use the node template to create node groups in any zone of this region.

5. Specify the Zone and click Continue.

6. In the Node template list, click Create node template to begin creating a sole-tenant node template.

7. Specify a Name for the node template.

8. Specify the Node type for each sole-tenant node in the node group to create based on this node template.

9. Optionally, you can also specify the following properties for the node template:

   • Add a Local SSD and GPU accelerator.
   • Select Enable CPU overcommit to control CPU overcommit levels for each VM scheduled on the node.
   • Add Node affinity labels. Affinity labels let you logically group nodes and node groups, and later, when provisioning VMs, you can specify affinity labels on the VMs to schedule VMs on a specific set of nodes or node groups. For more information, see Node affinity and anti-affinity.

10. Click Create to finish creating your node template.

11. Optional: to add a new sole-tenant node template in a different region, repeat the preceding steps.

To view the node templates, click Node templates in the Sole-tenant nodes page.

gcloud

Use the gcloud compute sole-tenancy node-templates create command to create a node template:

gcloud compute sole-tenancy node-templates create TEMPLATE_NAME \
  --node-type=NODE_TYPE \
  [--region=REGION \]
  [--node-affinity-labels=AFFINITY_LABELS \]
  [--accelerator type=GPU_TYPE,count=GPU_COUNT \]
  [--disk type=local-ssd,count=DISK_COUNT,size=DISK_SIZE \]
  [--cpu-overcommit-type=CPU_OVERCOMMIT_TYPE]

Replace the following:

• TEMPLATE_NAME: the name for the new node template.

• NODE_TYPE: the node type for sole-tenant nodes created based on this template. Use the gcloud compute sole-tenancy node-types list command to get a list of the node types available in each zone.

• REGION: the region to create the node template in. You can use this template to create node groups in any zone of this region.

• AFFINITY_LABELS: the keys and values, [KEY=VALUE,...], for affinity labels. Affinity labels let you logically group nodes and node groups and later, when provisioning VMs, you can specify affinity labels on the VMs to schedule VMs on a specific set of nodes or node groups. For more information, see Node affinity and anti-affinity.

• GPU_TYPE: the type of GPU for each sole-tenant node created based on this node template. For information on the zonal availability of GPUs, use the gcloud compute accelerator-types list command and choose a zone where either the n1, g2, a3-highgpu, or a3-megagpu sole-tenant node type is available. Depending on the zonal availability, set to one of:

  • For A3 High: nvidia-h100-80gb
  • For A3 Mega: nvidia-h100-mega-80gb
  • For G2: nvidia-l4
  • For N1:
    • nvidia-tesla-p100
    • nvidia-tesla-p4
    • nvidia-tesla-t4
    • nvidia-tesla-v100

• GPU_COUNT: the number of GPUs to specify depending on the type of GPU. Set to the value specified for the type of GPU as shown in the following table:

  GPU_TYPE	GPU_COUNT
  nvidia-h100-80gb	8
  nvidia-h100-mega-80gb	8
  nvidia-l4	8
  nvidia-tesla-p100	4
  nvidia-tesla-p4	4
  nvidia-tesla-t4	4
  nvidia-tesla-v100	8

• DISK_COUNT: number of Local SSD disks. Set to 16 or 24. This parameter is not required for A3 High and A3 Mega node types because they include a fixed number of Local SSD disks.

• DISK_SIZE: optional value for the partition size of the Local SSD in GB. The only supported partition size is 375, and if you don't set this value it defaults to 375.

• CPU_OVERCOMMIT_TYPE: the overcommit type for CPUs on a VM. Set to enabled or none.

REST

Use the nodeTemplates.insert method to create a node template:

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/nodeTemplates

{
  "name": "TEMPLATE_NAME",
  "nodeType": "NODE_TYPE",
  "nodeAffinityLabels": {
    "KEY": "VALUE",
    ...
  },
  "accelerators": [
    {
      "acceleratorType": "GPU_TYPE",
      "acceleratorCount": GPU_COUNT
    }
  ],
  "disks": [
    {
      "diskType": "local-ssd",
      "diskSizeGb": DISK_SIZE,
      "diskCount": DISK_COUNT
    }
  ],
  "cpuOvercommitType": CPU_OVERCOMMIT_TYPE
}

Replace the following:

• PROJECT_ID: the project ID.

• REGION: the region to create the node template in. You can use this template to create node groups in any zone of this region.

• TEMPLATE_NAME: the name for the new node template.

• NODE_TYPE: the node type for sole-tenant nodes created based on this template. Use the nodeTypes.list method to get a list of the node types available in each zone.

• KEY: the nodeAffinityLabels value that specifies the key portion of a node affinity label expressed as a key-value pair. Affinity labels let you logically group nodes and node groups, and later, when provisioning VMs, you can specify affinity labels on the VMs to schedule VMs on a specific set of nodes or node groups. For more information, see Node affinity and anti-affinity.

• VALUE: the nodeAffinityLabels value that specifies the value portion of a node affinity label key-value pair.

• GPU_TYPE: the type of GPU for each sole-tenant node created based on this node template. For information on the zonal availability of GPUs, use the gcloud compute accelerator-types list command and choose a zone where either the n1, g2, a3-highgpu, or a3-megagpu sole-tenant node type is available. Depending on the zonal availability, set to one of:

  • For A3 High: nvidia-h100-80gb
  • For A3 Mega: nvidia-h100-mega-80gb
  • For G2: nvidia-l4
  • For N1:
    • nvidia-tesla-p100
    • nvidia-tesla-p4
    • nvidia-tesla-t4
    • nvidia-tesla-v100

• GPU_COUNT: the number of GPUs for each sole-tenant node created based on this node template. Set to the value specified for the type of GPU as shown in the following table:

  GPU_TYPE	GPU_COUNT
  nvidia-h100-80gb	8
  nvidia-h100-mega-80gb	8
  nvidia-l4	8
  nvidia-tesla-p100	4
  nvidia-tesla-p4	4
  nvidia-tesla-t4	4
  nvidia-tesla-v100	8

• DISK_SIZE: optional value for the partition size of the Local SSD in GB. The only supported partition size is 375, and if you don't set this value it defaults to 375.

• DISK_COUNT: number of Local SSD disks. Set to 16 or 24. This parameter is not required for A3 High and A3 Mega node types because they include a fixed number of Local SSD disks.

• CPU_OVERCOMMIT_TYPE: CPU overcommit type. Set to enabled, none, or CPU_OVERCOMMIT_TYPE_UNSPECIFIED.

Console

1. In the Google Cloud console, go to the Sole-tenant nodes page.

   Go to Sole-tenant nodes

2. Click Node templates.

3. Select the name of an unused node template.

4. Click Delete.

gcloud

Use the gcloud compute sole-tenancy node-templates delete command to delete an unused node template:

gcloud compute sole-tenancy node-templates delete TEMPLATE_NAME \
  --region=REGION

Replace the following:

• TEMPLATE_NAME: the name of the node template to delete

• REGION: the region of the node template

REST

Use the compute.nodeTemplates.delete method to delete an unused node template:

 DELETE https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/nodeTemplates/TEMPLATE_NAME
 

Replace the following:

• PROJECT_ID: your project ID

• REGION: the Google Cloud region that contains the node template

• TEMPLATE_NAME: the name of the node template to delete