Compute Cloud API, gRPC: GpuClusterService

A set of methods for managing GPU clusters.

Call	Description
Get	Returns the specified GPU cluster.
List	Retrieves the list of GPU clusters in the specified folder.
Create	Creates a GPU cluster in the specified folder.
Update	Updates the specified GPU cluster.
Delete	Deletes the specified GPU cluster.
ListOperations	Lists operations for the specified GPU cluster.
ListInstances	List instances created in this GPU cluster.
ListAccessBindings	Lists access bindings for the GPU cluster.
SetAccessBindings	Sets access bindings for the GPU cluster.
UpdateAccessBindings	Updates access bindings for the GPU cluster.

Calls GpuClusterServiceCalls GpuClusterService

GetGet

Returns the specified GPU cluster.
To get the list of available GPU clusters, make a List request.

rpc Get (GetGpuClusterRequest) returns (GpuCluster)

GetGpuClusterRequestGetGpuClusterRequest

Field	Description
gpu_cluster_id	string ID of the GPU cluster to return. To get a GPU cluster ID, make a GpuClusterService.List request.

GpuClusterGpuCluster

Field	Description
id	string ID of GPU cluster.
folder_id	string ID of the folder that the GPU cluster belongs to.
created_at	google.protobuf.Timestamp Creation timestamp.
name	string Name of the GPU cluster. The name is unique within the folder.
description	string Description of the GPU cluster.
labels	map<string,string> GPU cluster labels as key:value pairs.
status	enum Status Status of the GPU cluster. CREATING: GPU cluster is being created. READY: GPU cluster is ready to use. ERROR: GPU cluster encountered a problem and cannot operate. DELETING: GPU cluster is being deleted.
zone_id	string ID of the availability zone where the GPU cluster resides.
interconnect_type	enum GpuInterconnectType Type of interconnect used for this GPU cluster. INFINIBAND: InfiniBand interconnect.

ListList

Retrieves the list of GPU clusters in the specified folder.

rpc List (ListGpuClustersRequest) returns (ListGpuClustersResponse)

ListGpuClustersRequestListGpuClustersRequest

Field	Description
folder_id	string Required. ID of the folder to list GPU clusters in. To get the folder ID, make a yandex.cloud.resourcemanager.v1.FolderService.List request. The maximum string length in characters is 50.
page_size	int64 The maximum number of results per page to return. If the number of available results is larger than page_size, the service returns a ListGpuClustersResponse.next_page_token that can be used to get the next page of results in subsequent list requests. The maximum value is 1000.
page_token	string Page token. To get the next page of results, set page_token to the ListGpuClustersResponse.next_page_token returned by a previous list request. The maximum string length in characters is 100.
filter	string A filter expression that filters GPU clusters listed in the response. The expression must specify: The field name. Currently you can use filtering only on GpuCluster.name field. An operator. Can be either = or != for single values, IN or NOT IN for lists of values. The value. Must be 3-63 characters long and match the regular expression ^[a-z][-a-z0-9]{1,61}[a-z0-9]. Example of a filter: name=my-schedule. The maximum string length in characters is 1000.
order_by	string A sorting expression that sorts GPU clusters listed in the response. The expression must specify the field name from GpuCluster and ascending or descending order, e.g. createdAt desc. Default value: id asc. The maximum string length in characters is 100.

ListGpuClustersResponseListGpuClustersResponse

Field	Description
gpu_clusters[]	GpuCluster List of GPU clusters in the specified folder.
next_page_token	string Token for getting the next page of the list. If the number of results is greater than the specified ListGpuClustersRequest.page_size, use next_page_token as the value for the ListGpuClustersRequest.page_token parameter in the next list request. Each subsequent page will have its own next_page_token to continue paging through the results.

GpuClusterGpuCluster

Field	Description
id	string ID of GPU cluster.
folder_id	string ID of the folder that the GPU cluster belongs to.
created_at	google.protobuf.Timestamp Creation timestamp.
name	string Name of the GPU cluster. The name is unique within the folder.
description	string Description of the GPU cluster.
labels	map<string,string> GPU cluster labels as key:value pairs.
status	enum Status Status of the GPU cluster. CREATING: GPU cluster is being created. READY: GPU cluster is ready to use. ERROR: GPU cluster encountered a problem and cannot operate. DELETING: GPU cluster is being deleted.
zone_id	string ID of the availability zone where the GPU cluster resides.
interconnect_type	enum GpuInterconnectType Type of interconnect used for this GPU cluster. INFINIBAND: InfiniBand interconnect.

CreateCreate

Creates a GPU cluster in the specified folder.

rpc Create (CreateGpuClusterRequest) returns (operation.Operation)

Metadata and response of Operation:

    Operation.metadata:CreateGpuClusterMetadata

    Operation.response:GpuCluster

CreateGpuClusterRequestCreateGpuClusterRequest

Field	Description
folder_id	string ID of the folder to create a GPU cluster in. To get a folder ID, make a yandex.cloud.resourcemanager.v1.FolderService.List request.
name	string Name of the GPU cluster. The name must be unique within the folder.
description	string Description of the GPU cluster.
labels	map<string,string> GPU cluster labels as key:value pairs.
zone_id	string ID of the availability zone where the GPU cluster resides. To get a list of available zones use the yandex.cloud.compute.v1.ZoneService.List request.
interconnect_type	enum GpuInterconnectType Type of interconnect to use for this GPU cluster. INFINIBAND: InfiniBand interconnect.

OperationOperation

Field	Description
id	string ID of the operation.
description	string Description of the operation. 0-256 characters long.
created_at	google.protobuf.Timestamp Creation timestamp.
created_by	string ID of the user or service account who initiated the operation.
modified_at	google.protobuf.Timestamp The time when the Operation resource was last modified.
done	bool If the value is false, it means the operation is still in progress. If true, the operation is completed, and either error or response is available.
metadata	google.protobuf.Any<CreateGpuClusterMetadata> Service-specific metadata associated with the operation. It typically contains the ID of the target resource that the operation is performed on. Any method that returns a long-running operation should document the metadata type, if any.
result	oneof: error or response The operation result. If done == false and there was no failure detected, neither error nor response is set. If done == false and there was a failure detected, error is set. If done == true, exactly one of error or response is set.
error	google.rpc.Status The error result of the operation in case of failure or cancellation.
response	google.protobuf.Any<GpuCluster> if operation finished successfully.

CreateGpuClusterMetadataCreateGpuClusterMetadata

Field	Description
gpu_cluster_id	string ID of the GPU cluster that is being created.

GpuClusterGpuCluster

Field	Description
id	string ID of GPU cluster.
folder_id	string ID of the folder that the GPU cluster belongs to.
created_at	google.protobuf.Timestamp Creation timestamp.
name	string Name of the GPU cluster. The name is unique within the folder.
description	string Description of the GPU cluster.
labels	map<string,string> GPU cluster labels as key:value pairs.
status	enum Status Status of the GPU cluster. CREATING: GPU cluster is being created. READY: GPU cluster is ready to use. ERROR: GPU cluster encountered a problem and cannot operate. DELETING: GPU cluster is being deleted.
zone_id	string ID of the availability zone where the GPU cluster resides.
interconnect_type	enum GpuInterconnectType Type of interconnect used for this GPU cluster. INFINIBAND: InfiniBand interconnect.

UpdateUpdate

Updates the specified GPU cluster.
Currently only name, description and labels can be updated.

rpc Update (UpdateGpuClusterRequest) returns (operation.Operation)

Metadata and response of Operation:

    Operation.metadata:UpdateGpuClusterMetadata

    Operation.response:GpuCluster

UpdateGpuClusterRequestUpdateGpuClusterRequest

Field	Description
gpu_cluster_id	string ID of the GPU cluster to update. To get the GPU cluster ID, make a GpuClusterService.List request.
update_mask	google.protobuf.FieldMask Field mask that specifies which attributes of the GPU cluster should be updated.
name	string New name for the GPU cluster. The name must be unique within the folder.
description	string New description of the GPU cluster.
labels	map<string,string> New GPU cluster labels as key:value pairs. Existing set of labels is completely replaced by the provided set, so if you just want to add or remove a label: Get the current set of labels with a GpuClusterService.Get request. Add or remove a label in this set. Send the new set in this field.

OperationOperation

Field	Description
id	string ID of the operation.
description	string Description of the operation. 0-256 characters long.
created_at	google.protobuf.Timestamp Creation timestamp.
created_by	string ID of the user or service account who initiated the operation.
modified_at	google.protobuf.Timestamp The time when the Operation resource was last modified.
done	bool If the value is false, it means the operation is still in progress. If true, the operation is completed, and either error or response is available.
metadata	google.protobuf.Any<UpdateGpuClusterMetadata> Service-specific metadata associated with the operation. It typically contains the ID of the target resource that the operation is performed on. Any method that returns a long-running operation should document the metadata type, if any.
result	oneof: error or response The operation result. If done == false and there was no failure detected, neither error nor response is set. If done == false and there was a failure detected, error is set. If done == true, exactly one of error or response is set.
error	google.rpc.Status The error result of the operation in case of failure or cancellation.
response	google.protobuf.Any<GpuCluster> if operation finished successfully.

UpdateGpuClusterMetadataUpdateGpuClusterMetadata

Field	Description
gpu_cluster_id	string ID of the GPU cluster that is being updated.

GpuClusterGpuCluster

Field	Description
id	string ID of GPU cluster.
folder_id	string ID of the folder that the GPU cluster belongs to.
created_at	google.protobuf.Timestamp Creation timestamp.
name	string Name of the GPU cluster. The name is unique within the folder.
description	string Description of the GPU cluster.
labels	map<string,string> GPU cluster labels as key:value pairs.
status	enum Status Status of the GPU cluster. CREATING: GPU cluster is being created. READY: GPU cluster is ready to use. ERROR: GPU cluster encountered a problem and cannot operate. DELETING: GPU cluster is being deleted.
zone_id	string ID of the availability zone where the GPU cluster resides.
interconnect_type	enum GpuInterconnectType Type of interconnect used for this GPU cluster. INFINIBAND: InfiniBand interconnect.

DeleteDelete

Deletes the specified GPU cluster.
GPU cluster can be deleted only if it doesn't have any instances associated with it.

rpc Delete (DeleteGpuClusterRequest) returns (operation.Operation)

Metadata and response of Operation:

    Operation.metadata:DeleteGpuClusterMetadata

    Operation.response:google.protobuf.Empty

DeleteGpuClusterRequestDeleteGpuClusterRequest

Field	Description
gpu_cluster_id	string ID of the GPU cluster to delete. To get a GPU cluster ID, make a GpuClusterService.List request.

OperationOperation

Field	Description
id	string ID of the operation.
description	string Description of the operation. 0-256 characters long.
created_at	google.protobuf.Timestamp Creation timestamp.
created_by	string ID of the user or service account who initiated the operation.
modified_at	google.protobuf.Timestamp The time when the Operation resource was last modified.
done	bool If the value is false, it means the operation is still in progress. If true, the operation is completed, and either error or response is available.
metadata	google.protobuf.Any<DeleteGpuClusterMetadata> Service-specific metadata associated with the operation. It typically contains the ID of the target resource that the operation is performed on. Any method that returns a long-running operation should document the metadata type, if any.
result	oneof: error or response The operation result. If done == false and there was no failure detected, neither error nor response is set. If done == false and there was a failure detected, error is set. If done == true, exactly one of error or response is set.
error	google.rpc.Status The error result of the operation in case of failure or cancellation.
response	google.protobuf.Any<google.protobuf.Empty> if operation finished successfully.

DeleteGpuClusterMetadataDeleteGpuClusterMetadata

Field	Description
gpu_cluster_id	string ID of the GPU cluster that is being deleted.

ListOperationsListOperations

Lists operations for the specified GPU cluster.

rpc ListOperations (ListGpuClusterOperationsRequest) returns (ListGpuClusterOperationsResponse)

ListGpuClusterOperationsRequestListGpuClusterOperationsRequest

Field	Description
gpu_cluster_id	string ID of the GPU cluster to list operations for. To get a GPU cluster ID, make a GpuClusterService.List request.
page_size	int64 The maximum number of results per page to return. If the number of available results is larger than page_size, the service returns a ListGpuClusterOperationsResponse.next_page_token that can be used to get the next page of results in subsequent list requests. Default value: 100.
page_token	string Page token. To get the next page of results, set page_token to the ListGpuClusterOperationsResponse.next_page_token returned by a previous list request.

ListGpuClusterOperationsResponseListGpuClusterOperationsResponse

Field	Description
operations[]	operation.Operation List of operations for the specified GPU cluster.
next_page_token	string Token for getting the next page of the list. If the number of results is greater than the specified ListGpuClusterOperationsRequest.page_size, use next_page_token as the value for the ListGpuClusterOperationsRequest.page_token parameter in the next list request. Each subsequent page will have its own next_page_token to continue paging through the results.

OperationOperation

Field	Description
id	string ID of the operation.
description	string Description of the operation. 0-256 characters long.
created_at	google.protobuf.Timestamp Creation timestamp.
created_by	string ID of the user or service account who initiated the operation.
modified_at	google.protobuf.Timestamp The time when the Operation resource was last modified.
done	bool If the value is false, it means the operation is still in progress. If true, the operation is completed, and either error or response is available.
metadata	google.protobuf.Any Service-specific metadata associated with the operation. It typically contains the ID of the target resource that the operation is performed on. Any method that returns a long-running operation should document the metadata type, if any.
result	oneof: error or response The operation result. If done == false and there was no failure detected, neither error nor response is set. If done == false and there was a failure detected, error is set. If done == true, exactly one of error or response is set.
error	google.rpc.Status The error result of the operation in case of failure or cancellation.
response	google.protobuf.Any The normal response of the operation in case of success. If the original method returns no data on success, such as Delete, the response is google.protobuf.Empty. If the original method is the standard Create/Update, the response should be the target resource of the operation. Any method that returns a long-running operation should document the response type, if any.

ListInstancesListInstances

List instances created in this GPU cluster.

rpc ListInstances (ListGpuClusterInstancesRequest) returns (ListGpuClusterInstancesResponse)

ListGpuClusterInstancesRequestListGpuClusterInstancesRequest

Field	Description
gpu_cluster_id	string ID of the GPU cluster to list instances in. To get a GPU cluster ID, make a GpuClusterService.List request.
page_size	int64 The maximum number of results per page to return. If the number of available results is larger than page_size, the service returns a ListGpuClusterInstancesResponse.next_page_token that can be used to get the next page of results in subsequent list requests. Default value: 100.
page_token	string Page token. To get the next page of results, set page_token to the ListGpuClusterInstancesResponse.next_page_token returned by a previous list request.
filter	string A filter expression that filters resources listed in the response. Currently you can use filtering only on the Instance.name field.

ListGpuClusterInstancesResponseListGpuClusterInstancesResponse

Field	Description
instances[]	Instance List of instances in the specified GPU cluster.
next_page_token	string Token for getting the next page of the list. If the number of results is greater than the specified ListGpuClusterInstancesRequest.page_size, use next_page_token as the value for the ListGpuClusterInstancesRequest.page_token parameter in the next list request. Each subsequent page will have its own next_page_token to continue paging through the results.

InstanceInstance

Field	Description
id	string ID of the instance.
folder_id	string ID of the folder that the instance belongs to.
created_at	google.protobuf.Timestamp
name	string Name of the instance. 1-63 characters long.
description	string Description of the instance. 0-256 characters long.
labels	map<string,string> Resource labels as key:value pairs. Maximum of 64 per resource.
zone_id	string ID of the availability zone where the instance resides.
platform_id	string ID of the hardware platform configuration for the instance.
resources	Resources Computing resources of the instance such as the amount of memory and number of cores.
status	enum Status Status of the instance. PROVISIONING: Instance is waiting for resources to be allocated. RUNNING: Instance is running normally. STOPPING: Instance is being stopped. STOPPED: Instance stopped. STARTING: Instance is being started. RESTARTING: Instance is being restarted. UPDATING: Instance is being updated. ERROR: Instance encountered a problem and cannot operate. CRASHED: Instance crashed and will be restarted automatically. DELETING: Instance is being deleted.
metadata	map<string,string> The metadata key:value pairs assigned to this instance. This includes custom metadata and predefined keys. For example, you may use the metadata in order to provide your public SSH key to the instance. For more information, see Metadata.
metadata_options	MetadataOptions Options allow user to configure access to instance's metadata
boot_disk	AttachedDisk Boot disk that is attached to the instance.
secondary_disks[]	AttachedDisk Array of secondary disks that are attached to the instance.
local_disks[]	AttachedLocalDisk Array of local disks that are attached to the instance.
filesystems[]	AttachedFilesystem Array of filesystems that are attached to the instance.
network_interfaces[]	NetworkInterface Array of network interfaces that are attached to the instance.
serial_port_settings	SerialPortSettings Serial port settings
gpu_settings	GpuSettings GPU settings
fqdn	string A domain name of the instance. FQDN is defined by the server in the format <hostname>.<region_id>.internal when the instance is created. If the hostname were not specified when the instance was created, FQDN would be <id>.auto.internal.
scheduling_policy	SchedulingPolicy Scheduling policy configuration.
service_account_id	string ID of the service account to use for authentication inside the instance. To get the service account ID, use a yandex.cloud.iam.v1.ServiceAccountService.List request.
network_settings	NetworkSettings Network Settings
placement_policy	PlacementPolicy Placement policy configuration.
host_group_id	string ID of the dedicated host group that the instance belongs to.
host_id	string ID of the dedicated host that the instance belongs to.
maintenance_policy	enum MaintenancePolicy Behaviour on maintenance events RESTART: Restart instance to move it to another host during maintenance MIGRATE: Use live migration to move instance to another host during maintenance
maintenance_grace_period	google.protobuf.Duration Time between notification via metadata service and maintenance
hardware_generation	HardwareGeneration This feature set is inherited from the image/disk used as a boot one at the creation of the instance.

ResourcesResources

Field	Description
memory	int64 The amount of memory available to the instance, specified in bytes.
cores	int64 The number of cores available to the instance.
core_fraction	int64 Baseline level of CPU performance with the ability to burst performance above that baseline level. This field sets baseline performance for each core.
gpus	int64 The number of GPUs available to the instance.

MetadataOptionsMetadataOptions

Field	Description
gce_http_endpoint	enum MetadataOption Enabled access to GCE flavored metadata ENABLED: Option is enabled DISABLED: Option is disabled
aws_v1_http_endpoint	enum MetadataOption Enabled access to AWS flavored metadata (IMDSv1) ENABLED: Option is enabled DISABLED: Option is disabled
gce_http_token	enum MetadataOption Enabled access to IAM credentials with GCE flavored metadata ENABLED: Option is enabled DISABLED: Option is disabled
aws_v1_http_token	enum MetadataOption Enabled access to IAM credentials with AWS flavored metadata (IMDSv1) ENABLED: Option is enabled DISABLED: Option is disabled

AttachedDiskAttachedDisk

Field	Description
mode	enum Mode Access mode to the Disk resource. READ_ONLY: Read-only access. READ_WRITE: Read/Write access.
device_name	string Serial number that is reflected into the /dev/disk/by-id/ tree of a Linux operating system running within the instance. This value can be used to reference the device for mounting, resizing, and so on, from within the instance.
auto_delete	bool Specifies whether the disk will be auto-deleted when the instance is deleted.
disk_id	string ID of the disk that is attached to the instance.

AttachedLocalDiskAttachedLocalDisk

Field	Description
size	int64 Size of the disk, specified in bytes.
device_name	string Serial number that is reflected into the /dev/disk/by-id/ tree of a Linux operating system running within the instance. This value can be used to reference the device for mounting, resizing, and so on, from within the instance.

AttachedFilesystemAttachedFilesystem

Field	Description
mode	enum Mode Access mode to the filesystem. READ_ONLY: Read-only access. READ_WRITE: Read/Write access.
device_name	string Name of the device representing the filesystem on the instance. The name should be used for referencing the filesystem from within the instance when it's being mounted, resized etc.
filesystem_id	string ID of the filesystem that is attached to the instance.

NetworkInterfaceNetworkInterface

Field	Description
index	string The index of the network interface, will be generated by the server, 0,1,2... etc if not specified.
mac_address	string MAC address that is assigned to the network interface.
subnet_id	string ID of the subnet.
primary_v4_address	PrimaryAddress Primary IPv4 address that is assigned to the instance for this network interface.
primary_v6_address	PrimaryAddress Primary IPv6 address that is assigned to the instance for this network interface. IPv6 not available yet.
security_group_ids[]	string ID's of security groups attached to the interface

PrimaryAddressPrimaryAddress

Field	Description
address	string An IPv4 internal network address that is assigned to the instance for this network interface.
one_to_one_nat	OneToOneNat One-to-one NAT configuration. If missing, NAT has not been set up.
dns_records[]	DnsRecord Internal DNS configuration

OneToOneNatOneToOneNat

Field	Description
address	string An external IP address associated with this instance.
ip_version	enum IpVersion IP version for the external IP address. IPV4: IPv4 address, for example 192.0.2.235. IPV6: IPv6 address. Not available yet.
dns_records[]	DnsRecord External DNS configuration

DnsRecordDnsRecord

Field	Description
fqdn	string Name of the A/AAAA record as specified when creating the instance. Note that if `fqdn' has no trailing '.', it is specified relative to the zone (@see dns_zone_id).
dns_zone_id	string DNS zone id for the record (optional, if not set, some private zone is used).
ttl	int64 DNS record ttl (optional, if not set, a reasonable default is used.)
ptr	bool When true, indicates there is a corresponding auto-created PTR DNS record.

SerialPortSettingsSerialPortSettings

Field	Description
ssh_authorization	enum SSHAuthorization Authentication and authorization in serial console when using SSH protocol INSTANCE_METADATA: Authentication and authorization using SSH keys in instance metadata OS_LOGIN: Authentication and authorization using Oslogin service

GpuSettingsGpuSettings

Field	Description
gpu_cluster_id	string Attach instance to specified GPU cluster.

SchedulingPolicySchedulingPolicy

Field	Description
preemptible	bool True for short-lived compute instances. For more information, see Preemptible VMs.

NetworkSettingsNetworkSettings

Field	Description
type	enum Type Network Type STANDARD: Standard network. SOFTWARE_ACCELERATED: Software accelerated network. HARDWARE_ACCELERATED: Hardware accelerated network (not available yet, reserved for future use).

PlacementPolicyPlacementPolicy

Field	Description
placement_group_id	string Placement group ID.
host_affinity_rules[]	HostAffinityRule List of affinity rules. Scheduler will attempt to allocate instances according to order of rules.
placement_group_partition	int64 Placement group partition

HostAffinityRuleHostAffinityRule

Field	Description
key	string Affinity label or one of reserved values - 'yc.hostId', 'yc.hostGroupId'
op	enum Operator Include or exclude action
values[]	string Affinity value or host ID or host group ID

HardwareGenerationHardwareGeneration

Field	Description
features	oneof: legacy_features or generation2_features
legacy_features	LegacyHardwareFeatures
generation2_features	Generation2HardwareFeatures

LegacyHardwareFeaturesLegacyHardwareFeatures

Field	Description
pci_topology	enum PCITopology

Generation2HardwareFeaturesGeneration2HardwareFeatures

Empty.

ListAccessBindingsListAccessBindings

Lists access bindings for the GPU cluster.

rpc ListAccessBindings (ListAccessBindingsRequest) returns (ListAccessBindingsResponse)

ListAccessBindingsRequestListAccessBindingsRequest

Field	Description
resource_id	string Required. ID of the resource to list access bindings for. To get the resource ID, use a corresponding List request. For example, use the yandex.cloud.resourcemanager.v1.CloudService.List request to get the Cloud resource ID. The maximum string length in characters is 50.
page_size	int64 The maximum number of results per page that should be returned. If the number of available results is larger than page_size, the service returns a ListAccessBindingsResponse.next_page_token that can be used to get the next page of results in subsequent list requests. Default value: 100. The maximum value is 1000.
page_token	string Page token. Set page_token to the ListAccessBindingsResponse.next_page_token returned by a previous list request to get the next page of results. The maximum string length in characters is 100.

ListAccessBindingsResponseListAccessBindingsResponse

Field	Description
access_bindings[]	AccessBinding List of access bindings for the specified resource.
next_page_token	string This token allows you to get the next page of results for list requests. If the number of results is larger than ListAccessBindingsRequest.page_size, use the next_page_token as the value for the ListAccessBindingsRequest.page_token query parameter in the next list request. Each subsequent list request will have its own next_page_token to continue paging through the results.

AccessBindingAccessBinding

Field	Description
role_id	string Required. ID of the yandex.cloud.iam.v1.Role that is assigned to the subject. The maximum string length in characters is 50.
subject	Subject Required. Identity for which access binding is being created. It can represent an account with a unique ID or several accounts with a system identifier.

SubjectSubject

Field	Description
id	string Required. ID of the subject. It can contain one of the following values: allAuthenticatedUsers: A special system identifier that represents anyone who is authenticated. It can be used only if the type is system. allUsers: A special system identifier that represents anyone. No authentication is required. For example, you don't need to specify the IAM token in an API query. <cloud generated id>: An identifier that represents a user account. It can be used only if the type is userAccount, federatedUser or serviceAccount. The maximum string length in characters is 50.
type	string Required. Type of the subject. It can contain one of the following values: userAccount: An account on Yandex or Yandex.Connect, added to Yandex.Cloud. serviceAccount: A service account. This type represents the yandex.cloud.iam.v1.ServiceAccount resource. federatedUser: A federated account. This type represents a user from an identity federation, like Active Directory. system: System group. This type represents several accounts with a common system identifier. For more information, see Subject to which the role is assigned. The maximum string length in characters is 100.

SetAccessBindingsSetAccessBindings

Sets access bindings for the GPU cluster.

rpc SetAccessBindings (SetAccessBindingsRequest) returns (operation.Operation)

Metadata and response of Operation:

    Operation.metadata:SetAccessBindingsMetadata

    Operation.response:AccessBindingsOperationResult

SetAccessBindingsRequestSetAccessBindingsRequest

Field	Description
resource_id	string Required. ID of the resource for which access bindings are being set. To get the resource ID, use a corresponding List request. The maximum string length in characters is 50.
access_bindings[]	AccessBinding Required. Access bindings to be set. For more information, see Access Bindings.

AccessBindingAccessBinding

Field	Description
role_id	string Required. ID of the yandex.cloud.iam.v1.Role that is assigned to the subject. The maximum string length in characters is 50.
subject	Subject Required. Identity for which access binding is being created. It can represent an account with a unique ID or several accounts with a system identifier.

SubjectSubject

Field	Description
id	string Required. ID of the subject. It can contain one of the following values: allAuthenticatedUsers: A special system identifier that represents anyone who is authenticated. It can be used only if the type is system. allUsers: A special system identifier that represents anyone. No authentication is required. For example, you don't need to specify the IAM token in an API query. <cloud generated id>: An identifier that represents a user account. It can be used only if the type is userAccount, federatedUser or serviceAccount. The maximum string length in characters is 50.
type	string Required. Type of the subject. It can contain one of the following values: userAccount: An account on Yandex or Yandex.Connect, added to Yandex.Cloud. serviceAccount: A service account. This type represents the yandex.cloud.iam.v1.ServiceAccount resource. federatedUser: A federated account. This type represents a user from an identity federation, like Active Directory. system: System group. This type represents several accounts with a common system identifier. For more information, see Subject to which the role is assigned. The maximum string length in characters is 100.

OperationOperation

Field	Description
id	string ID of the operation.
description	string Description of the operation. 0-256 characters long.
created_at	google.protobuf.Timestamp Creation timestamp.
created_by	string ID of the user or service account who initiated the operation.
modified_at	google.protobuf.Timestamp The time when the Operation resource was last modified.
done	bool If the value is false, it means the operation is still in progress. If true, the operation is completed, and either error or response is available.
metadata	google.protobuf.Any<SetAccessBindingsMetadata> Service-specific metadata associated with the operation. It typically contains the ID of the target resource that the operation is performed on. Any method that returns a long-running operation should document the metadata type, if any.
result	oneof: error or response The operation result. If done == false and there was no failure detected, neither error nor response is set. If done == false and there was a failure detected, error is set. If done == true, exactly one of error or response is set.
error	google.rpc.Status The error result of the operation in case of failure or cancellation.
response	google.protobuf.Any<AccessBindingsOperationResult> if operation finished successfully.

SetAccessBindingsMetadataSetAccessBindingsMetadata

Field	Description
resource_id	string ID of the resource for which access bindings are being set.

UpdateAccessBindingsUpdateAccessBindings

Updates access bindings for the GPU cluster.

rpc UpdateAccessBindings (UpdateAccessBindingsRequest) returns (operation.Operation)

Metadata and response of Operation:

    Operation.metadata:UpdateAccessBindingsMetadata

    Operation.response:AccessBindingsOperationResult

UpdateAccessBindingsRequestUpdateAccessBindingsRequest

Field	Description
resource_id	string Required. ID of the resource for which access bindings are being updated. The maximum string length in characters is 50.
access_binding_deltas[]	AccessBindingDelta Required. Updates to access bindings. The number of elements must be greater than 0.

AccessBindingDeltaAccessBindingDelta

Field	Description
action	enum AccessBindingAction Required. The action that is being performed on an access binding. ADD: Addition of an access binding. REMOVE: Removal of an access binding.
access_binding	AccessBinding Required. Access binding. For more information, see Access Bindings.

AccessBindingAccessBinding

Field	Description
role_id	string Required. ID of the yandex.cloud.iam.v1.Role that is assigned to the subject. The maximum string length in characters is 50.
subject	Subject Required. Identity for which access binding is being created. It can represent an account with a unique ID or several accounts with a system identifier.

SubjectSubject

Field	Description
id	string Required. ID of the subject. It can contain one of the following values: allAuthenticatedUsers: A special system identifier that represents anyone who is authenticated. It can be used only if the type is system. allUsers: A special system identifier that represents anyone. No authentication is required. For example, you don't need to specify the IAM token in an API query. <cloud generated id>: An identifier that represents a user account. It can be used only if the type is userAccount, federatedUser or serviceAccount. The maximum string length in characters is 50.
type	string Required. Type of the subject. It can contain one of the following values: userAccount: An account on Yandex or Yandex.Connect, added to Yandex.Cloud. serviceAccount: A service account. This type represents the yandex.cloud.iam.v1.ServiceAccount resource. federatedUser: A federated account. This type represents a user from an identity federation, like Active Directory. system: System group. This type represents several accounts with a common system identifier. For more information, see Subject to which the role is assigned. The maximum string length in characters is 100.

OperationOperation

Field	Description
id	string ID of the operation.
description	string Description of the operation. 0-256 characters long.
created_at	google.protobuf.Timestamp Creation timestamp.
created_by	string ID of the user or service account who initiated the operation.
modified_at	google.protobuf.Timestamp The time when the Operation resource was last modified.
done	bool If the value is false, it means the operation is still in progress. If true, the operation is completed, and either error or response is available.
metadata	google.protobuf.Any<UpdateAccessBindingsMetadata> Service-specific metadata associated with the operation. It typically contains the ID of the target resource that the operation is performed on. Any method that returns a long-running operation should document the metadata type, if any.
result	oneof: error or response The operation result. If done == false and there was no failure detected, neither error nor response is set. If done == false and there was a failure detected, error is set. If done == true, exactly one of error or response is set.
error	google.rpc.Status The error result of the operation in case of failure or cancellation.
response	google.protobuf.Any<AccessBindingsOperationResult> if operation finished successfully.

UpdateAccessBindingsMetadataUpdateAccessBindingsMetadata

Field	Description
resource_id	string ID of the resource for which access bindings are being updated.