Documentation | Everywhere Inference › Clusters

API Reference

Everywhere Inference

everywhere_inference

Methods

Get Flavors For AI Cluster -> { count, results }

get/v1/ai/flavors/{project_id}/{region_id}

Get flavors available for a potential AI cluster

Deprecated List Images For AI GPU Cluster -> { count, results }

get/v1/ai/images/gpu/{project_id}/{region_id}

Deprecated. Retrieve available AI GPU images.

List Available Regions That Provide AI Cluster Support -> { count, results }

get/v1/ai/regions

List available regions that provide AI cluster support

Everywhere Inference

Clusters

everywhere_inference.clusters

Methods

Assign Security Group To AI Cluster ->

post/v1/ai/clusters/{project_id}/{region_id}/{cluster_id}/addsecuritygroup

Assign the security group to the AI cluster. To assign multiple security groups to all ports, use the NULL value for the port_id field

Attach Interface To AI Cluster Node -> { tasks }

post/v1/ai/clusters/{project_id}/{region_id}/{instance_id}/attach_interface

Attach interface to AI cluster node

Check Quota For AI Cluster Creation -> { baremetal_basic_count_limit, baremetal_basic_count_requested, baremetal_basic_count_usage, 126 more... }

post/v1/ai/clusters/{project_id}/{region_id}/check_limits

Check if regional quota is exceeded, if yes the number of additional quotas needed to create the specified AI cluster will be calculated

Delete AI Cluster -> { tasks }

delete/v2/ai/clusters/{project_id}/{region_id}/{cluster_id}

Delete AI cluster

Detach Interface From AI Cluster Node -> { tasks }

post/v1/ai/clusters/{project_id}/{region_id}/{instance_id}/detach_interface

Detach interface from AI cluster node

Get AI Cluster Node Console URL -> { remote_console }

get/v1/ai/clusters/{project_id}/{region_id}/{instance_id}/get_console

Get AI cluster node console URL

List AI Clusters -> OffsetPage<{ cluster_id, cluster_status, creator_task_id, 20 more... }>

get/v2/ai/clusters/{project_id}/{region_id}

List AI clusters

Un Assign Security Group From AI Cluster ->

post/v1/ai/clusters/{project_id}/{region_id}/{cluster_id}/delsecuritygroup

Un-assign the security group from the AI cluster. To un-assign multiple security groups to all ports, use the NULL value for the port_id field

Get AI Cluster -> { cluster_id, cluster_status, creator_task_id, 20 more... }

get/v2/ai/clusters/{project_id}/{region_id}/{cluster_id}

Get AI cluster

Parameters

project_id: number

region_id: number

cluster_id: string

Response fields

cluster_id: string

AI Cluster ID

cluster_status: "ACTIVE" | "ERROR" | "PENDING" | 1 more...

AI Cluster status

creator_task_id: string

Task that created this entity

task_id: string

Task ID associated with the cluster

cluster_metadata: unknown

Optional

Cluster metadata (simple key-value pairs)

cluster_metadata_detailed: unknown

Optional

Detailed cluster metadata

cluster_name: string

Optional

AI Cluster Name

created_at: string

Optional

Datetime when the cluster was created

flavor: string

Optional

Flavor ID (name)

image_id: string

Optional

Image ID

image_name: string

Optional

Image name

interfaces: Array<{ network_id, port_id, subnet_id, 1 more... }>

Optional

Networks managed by user and associated with the cluster

keypair_name: string

Optional

password: string

Optional

A password for a bare metal server. This parameter is used to set a password for the "Admin" user on a Windows instance, a default user or a new user on a Linux instance

poplar_servers: Array<{ addresses, blackhole_ports, creator_task_id, 21 more... }>

Optional

Poplar servers

project_id: number

Optional

Project ID

region: string

Optional

Region name

region_id: number

Optional

Region ID

security_groups: Array<{ network_id, port_id, security_groups }>

Optional

Security groups attached to the cluster

task_status: "CLUSTER_CLEAN_UP" | "CLUSTER_RESIZE" | "CLUSTER_RESUME" | 8 more...

Optional

Task status

user_data: string

Optional

String in base64 format. Must not be passed together with 'username' or 'password'. Examples of the user_data: https://cloudinit.readthedocs.io/en/latest/topics/examples.html

username: string

Optional

A name of a new user in the Linux instance. It may be passed with a 'password' parameter

volumes: Array<{ id, attachments, bootable, 19 more... }>

Optional

List of volumes attached to the cluster

Request example

curl https://api.gcore.com/cloud/v2/ai/clusters/$PROJECT_ID/$REGION_ID/$CLUSTER_ID \
    -H "APIKey: $CLOUD_API_API_KEY"

200Example

{
  "cluster_id": "d1e1500b-e2be-40aa-9a4b-cc493fa1af30",
  "cluster_status": "ACTIVE",
  "creator_task_id": "d1e1500b-e2be-40aa-9a4b-cc493fa1af30",
  "task_id": "d1e1500b-e2be-40aa-9a4b-cc493fa1af30",
  "cluster_name": "cluster-name",
  "created_at": "2022-10-10T18:30:33Z",
  "flavor": "g2a-ai-60-116-1100-v1pod-4",
  "image_id": "5621ebbb-3731-4dc1-a2da-4bfc181fb8a6",
  "image_name": "ubuntu-20.04-x64-poplar-ironic-1.18.0-2.6.0",
  "task_status": "CLUSTER_CLEAN_UP"
}

Everywhere Inference Clusters

GPU

everywhere_inference.clusters.gpu

Methods

Create AI GPU Cluster -> { tasks }

post/v1/ai/clusters/gpu/{project_id}/{region_id}

Create a new AI cluster.

Rebuild One Or Many Nodes From GPU Cluster -> { tasks }

post/v1/ai/clusters/gpu/{project_id}/{region_id}/{cluster_id}/rebuild

Rebuild one or many nodes from GPU cluster. All cluster nodes need to be provided to change the cluster image.

Remove Single Node From GPU Cluster -> { tasks }

delete/v1/ai/clusters/gpu/{project_id}/{region_id}/{cluster_id}/node/{instance_id}

Remove single node from GPU cluster.

Resize AI GPU Cluster -> { tasks }

post/v1/ai/clusters/gpu/{project_id}/{region_id}/{cluster_id}/resize

Resize an existing AI GPU cluster.

Everywhere Inference Clusters

Metadata

everywhere_inference.clusters.metadata

Methods

Create Or Update Metadata Of AI Cluster ->

post/v2/ai/clusters/{project_id}/{region_id}/{cluster_id}/metadata

Create or update one or more metadata items of the cluster. If an item does not exist, it gets created in the cluster metadata. If an item already exists, it's value is overwritten

List All Metadata Of AI Cluster -> { count, results }

get/v1/ai/clusters/{project_id}/{region_id}/{cluster_id}/metadata

List all metadata of AI cluster

List Network Interfaces For AI Cluster -> { count, results }

get/v1/ai/clusters/{project_id}/{region_id}/{cluster_id}/interfaces

List network interfaces attached to AI cluster poplar servers

List Network Ports For AI Cluster -> { count, results }

get/v1/ai/clusters/{project_id}/{region_id}/{cluster_id}/ports

List network ports of AI cluster poplar servers

Replace Metadata ->

put/v2/ai/clusters/{project_id}/{region_id}/{cluster_id}/metadata

All existing metadata is deleted and replaced with the metadata from the request

Deprecated Create Or Update Metadata Of AI Cluster ->

put/v1/ai/clusters/{project_id}/{region_id}/{cluster_id}/metadata

Create or update one or more metadata items of the cluster. If an item does not exist, it gets created in the cluster metadata. If an item already exists, it's value is overwritten

Everywhere Inference Clusters

Metadata Item

everywhere_inference.clusters.metadata_item

Methods

Deprecated Delete AI Cluster Metadata Item By Key ->

delete/v1/ai/cluster/{project_id}/{region_id}/{cluster_id}/metadata_item

Delete AI cluster metadata item by key

Deprecated Get AI Cluster Metadata Item By Key -> { key, read_only, value }

get/v1/ai/cluster/{project_id}/{region_id}/{cluster_id}/metadata_item

Get AI cluster metadata item by key

Everywhere Inference Clusters

Powercycle

everywhere_inference.clusters.powercycle

Methods

Powercycle All AI Cluster Nodes -> { count, results }

post/v2/ai/clusters/{project_id}/{region_id}/{cluster_id}/powercycle

Powercycle (stop and start) all AI cluster nodes, aka hard reboot

Powercycle One AI Cluster Node -> { addresses, blackhole_ports, creator_task_id, 21 more... }

post/v1/ai/clusters/{project_id}/{region_id}/{instance_id}/powercycle

Powercycle (stop and start) one AI cluster node, aka hard reboot

Everywhere Inference Clusters

Reboot

everywhere_inference.clusters.reboot

Methods

Reboot All AI Cluster Nodes -> { count, results }

post/v2/ai/clusters/{project_id}/{region_id}/{cluster_id}/reboot

Reboot all AI cluster nodes

Reboot One AI Cluster Node -> { addresses, blackhole_ports, creator_task_id, 21 more... }

post/v1/ai/clusters/{project_id}/{region_id}/{instance_id}/reboot

Reboot one AI cluster node