Help Center/ ModelArts/ API Reference/ Model Inference/ Service Deployment Instance Management/ Manual Service Scaling

Updated on 2026-09-01 GMT+08:00

Manual Service Scaling

Function

This API is used to scale model service instances. By calling this API, you can scale an existing service without adding a new version. Errors include permission verification errors, service status errors, and parameter verification errors.

Debugging

You can debug this API through automatic authentication in API Explorer or use the SDK sample code generated by API Explorer.

Authorization Information

Each account has all the permissions required to call all APIs, but IAM users must be assigned the required permissions.

If you are using role/policy-based authorization, see Permissions Policies and Supported Actions for details on the required permissions.
If you are using identity policy-based authorization, the following identity policy-based permissions are required.

Action

Access Level

Resource Type (*: required)

Condition Key

Alias

Dependencies

modelarts:service:scaleDeployment

Read

service *

-

-

-

Action	Access Level	Resource Type (*: required)	Condition Key	Alias	Dependencies
modelarts:service:scaleDeployment	Read	service *	-	-	-

URI

PUT /v2/{project_id}/services/{service_id}/deployments/{deployment_name}/scale

**Table 1** Path Parameters
Parameter	Mandatory	Type	Description
project_id	Yes	String	Definition Project ID. For details, see Obtaining a Project ID and Name. Constraints N/A Range Project ID. Default Value N/A
service_id	Yes	String	Definition Service ID
deployment_name	Yes	String	Definition Deployment name, which can be obtained from the response body during deployment creation. You can also obtain the deployments owned by the current user by obtaining service deployments. The name field indicates the deployment name. Constraints N/A Range N/A Default Value N/A

Request Parameters

**Table 2** Request header parameters
Parameter	Mandatory	Type	Description
X-Auth-Token	Yes	String	Definition User token. It can be obtained by calling the IAM API used to obtain a user token. The value of X-Subject-Token in the response header is the user token. For details about how to obtain the token, see Obtaining a User Token Through Password Authentication. Constraints N/A Range N/A Default Value N/A
Content-Type	Yes	String	Definition Message body type or format. Constraints N/A Range application/json application/json;charset=utf-8 Default Value N/A

**Table 3** Request body parameters
Parameter	Mandatory	Type	Description
count	Yes	Integer	Definition Number of service instances. Range [1, 128]

Response Parameters

Status code: 200

**Table 4** Response body parameters
Parameter	Type	Description
id	String	Definition Deployment ID, which you can obtain from the response body when adding the deployment or by obtaining service deployments. The deployment_id field indicates the deployment ID. Range Deployment ID.
name	String	Definition Service deployment name.
infer_name	String	Definition Deployment ID (deprecated field). Range N/A
create_at	String	Definition Creation time, which is automatically generated. Range Millisecond-level timestamp, which is a 13-digit number.
description	String	Definition Description. Range N/A
pool_id	String	Definition Dedicated resource pool ID. Range N/A
framework	String	Definition Algorithm framework. Range COMMON: common real-time service.
priority	String	Definition Service priority. Range [1, 3]
secret_type	String	Definition Secret type. You can select the secret type (dew or agency) to be used here. Range [dew: DEW secret.] (tag:hws,hws_hk,fcs) agency: temporary agency credential.
status	String	Definition Service deployment status
count	Integer	Definition Number of service instances.
high_avail_switch	String	Definition Specifies whether to enable HA. Once enabled, pods of different instances are evenly distributed to different nodes. Range N/A true
system_log_dump_enable	String	Definition Specifies whether to enable system log dump. Range N/A false
unit_configs	Array of UnitConfig objects	Definition Instance unit configuration.
update_at	String	Definition Update time, which is automatically generated. Range Millisecond-level timestamp, which is a 13-digit number.
version	String	Definition Current service version.
version_count	Integer	Definition Number of service versions. Range N/A
weight	Integer	Definition Weight percentage of the traffic allocated to the model. This parameter is mandatory only when the model is deployed as a real-time service. Range [0,100]
advanced_config	AdvancedConfig object	Definition Advanced service deployment settings.
job_id	String	Definition Wushan workflow ID. Range N/A
deployment_name	String	Definition Service deployment name.
frozen_infos	Array of FrozenInfo objects	Definition Freezing type returned when a service or deployment is frozen.

**Table 5** UnitConfig
Parameter	Type	Description
id	String	Definition Instance unit ID. Range N/A
name	String	Definition Role name of the instance unit.
role	String	Definition Instance unit role. Range COMMON: other roles
custom_spec	CustomResourceSpec object	Definition Custom resource specifications. Example "custom_spec": { "arch": "x86", "cpu": 0.51, "memory": 1024.0 },
flavor	String	Definition Resource specifications. Select specifications based on service requirements. When specification is set to custom, the custom specifications are used. custom_spec specifies deployment specifications. Range N/A
flavor_display_name	String	Definition Display name of a resource specification. Range N/A
image	ImageInfo object	Definition Image configuration information, including the image path and image source.
models	Array of ModelResource objects	Definition This parameter is discarded. You are advised to use files to configure models. Model configuration. You can select a model and weight file to use with the image.
codes	Array of Code objects	Definition This parameter is discarded. You are advised to use files to configure code. Code configuration. You can select the OBS path where the code is stored.
files	Array of FileInfo objects	Definition Model and code configuration. You can select the model and weight file to work with the image and the OBS path where the code is stored.
dumps	Array of Dump objects	Definition Dump configuration. You can select the OBS bucket where the data is to be dumped.
count	Integer	Definition Number of configuration instances. Range N/A
cmd	String	Definition Startup command. Range N/A
termination_grace	TerminationGrace object	Definition Graceful shutdown configuration.
envs	Map<String,String>	Definition Environment variable.
readiness_health	Health object	Definition Readiness health check.
startup_health	Health object	Definition Health checks are enabled.
liveness_health	Health object	Definition Liveness health check. Range N/A
port	Integer	Definition Port. Range [1, 65535]
recovery	String	Definition Auto-rebuild policy. When enabled, the platform will automatically rebuild pods according to the policy if they restart due to deployment configuration changes or hardware failures. If disabled, the platform will not intervene. Range Instance: Deployment replica rebuilding. Restarts the entire deployment when a fault occurs. Role: Unit rebuilding. Restarts all pods within a deployment unit if any pod in that unit fails. Pod: Pod rebuilding. Restarts the faulty pod if it fails.
npu_reset_enable	Boolean	Definition Specifies whether to enable the recovery policy. Range true: The recovery policy is enabled. false: The recovery policy is disabled.
group_count	Integer	Definition Number of replicas in a unit. This parameter is invalid when deploy_type is set to SINGLE or workload_type is set to DEPLOYMENT. Range [1, 100] or left empty. 1
affinity	Affinity object	Definition Node affinity scheduling.
security_config	ServiceSecurityConfig object	Definition Container user ID/user group ID.
pool_resource_flavor	String	Definition Node pool resource specifications. Range The value can contain a maximum of 128 characters.

**Table 6** CustomResourceSpec
Parameter	Type	Description
gpu	Float	Definition Number of GPUs. Range The value can be a decimal. It cannot be smaller than 0, with the third decimal place is rounded off.
memory	Integer	Definition Memory size, in MB. Range Only integers are supported.
cpu	Float	Definition CPU cores. Range The value can be a decimal. It cannot be smaller than 0.01, with the third decimal place is rounded off.
ascend	Integer	Definition Number of Ascend chips. Range N/A
arch	String	Definition Architecture type. Range x86/arm64.

**Table 7** ImageInfo
Parameter	Type	Description
id	String	Definition Image ID. Range N/A
source	String	Definition Image type, which indicates the image source. Range SWR: SoftWare Repository for Container (SWR) [IMAGE: [general image].] (tag:hws,hws_hk)
swr_path	String	Definition Image path. The address is different when the source value is different. Range N/A
category	String	Definition Flavors supported by the image. Range GPU: graphics processing unit. CPU: central processing unit. ASCEND: Ascend chip. CPU

**Table 8** ModelResource
Parameter	Type	Description
source	String	Definition Code source type. Range The value is case-insensitive. OBS: Object Storage Service OBSFS: OBS file system API EFS: Scalable File Service LOCAL: local storage directory to which the host machine is mounted.
address	String	Definition Code source address, which complies with the format of different storage systems. Range N/A
mount_path	String	Definition Path mounted to the container. The value must start with a slash (/) and can contain hyphens (-), backslashes (), underscores (), periods (.), letters, and digits. Range* The value must start and end with a slash (/) and can contain at most 255 characters. Only letters, digits, hyphens (-), and underscores (*) are allowed.
host_cache	Boolean	Definition Specifies whether the model can be cached locally. This function is disabled by default. Range true: supported. false: not supported. false
efs_sub_path	String	Definition When the storage type is EFS, subdirectories can be configured. Range N/A
read_only	Boolean	Definition Specifies whether the mount permission is read-only. Range true: read only false: non-read-only
os_warm_up	Boolean	Definition OS warmup. Range true: Warmup enabled. false: Warmup disabled.
source_name	String	Definition Warmup name. os_warm_up is set to true. Range The value can contain at most 64 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.
asset_id	String	Definition Preset asset ID. Range The value can contain at most 64 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.

**Table 9** Code
Parameter	Type	Description
source	String	Definition Code source type. Range OBS: Object Storage Service OBSFS: OBS file system API EFS: Scalable File Service
address	String	Definition Code source address, which complies with the format of different storage systems. Range N/A
source_id	String	Definition Code source ID. Either this parameter or address must be specified. The SFS Turbo ID can be input only when source is set to EFS. Range N/A
mount_path	String	Definition Path mounted to the container. Range The value must start and end with a slash (/) and can contain at most 255 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.
efs_sub_path	String	Definition EFS subpath. Range N/A
read_only	Boolean	Definition Specifies whether the mount permission is read-only. Range true: read only false: non-read-only

**Table 10** FileInfo
Parameter	Type	Description
source	String	Definition Code source type. Range The value is case-insensitive. OBS: Object Storage Service OBSFS: OBS file system API EFS: Scalable File Service LOCAL: local storage directory to which the host machine is mounted.
address	String	Definition Code source address, which complies with the format of different storage systems. Range N/A
mount_path	String	Definition Path mounted to the container. Range The value must start and end with a slash (/) and can contain at most 255 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.
host_cache	Boolean	Definition Specifies whether the model can be cached locally. This function is disabled by default. Range true: supported. false: not supported. false
efs_sub_path	String	Definition When the storage type is EFS, subdirectories can be configured. Range N/A
read_only	Boolean	Definition Specifies whether the mount permission is read-only. Range true: read only false: non-read-only
os_warm_up	Boolean	Definition OS warmup. Range true: Warmup enabled. false: Warmup disabled.
source_name	String	Definition Warmup name. os_warm_up is set to true. Range The value can contain at most 64 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.

**Table 11** Dump
Parameter	Type	Description
source	String	Definition Source type of the dump mount directory. Range OBS: Object Storage Service (tag:hws,hws_hk) OBSFS: OBS file system API
address	String	Definition Source address of the dump mount directory. The value can be an HTTPS or OBS address. Range N/A
mount_path	String	Definition Path mounted to the container. The value must start with a slash (/) and can contain hyphens (-), backslashes (), underscores (_), periods (.), letters, and digits. Range N/A

**Table 12** TerminationGrace
Parameter	Type	Description
pre_stop_cmd	String	Definition: The shutdown command is triggered when the container receives a stop signal, but it must be completed within the grace period or the shutdown times out. If this happens, the container will be forcefully stopped. You can use this command to perform operations such as closing database connections, releasing file handles, and stopping child processes. Range Protocol: http or https Port: 1 to 65535 Address: Only letters, digits, periods (.), hyphens (-), underscores (_), and slashes (/) are allowed. The value cannot start with a slash (/).
termination_grace_period_seconds	Integer	Definition This parameter indicates the maximum length of time that can pass between when a Pod receives a stop signal to when it is forcefully stopped. It is used for the Pod to perform cleanup operations (such as closing connections, releasing resources, and saving states). Range Protocol: http or https Port: 1 to 65535 Address: Only letters, digits, periods (.), hyphens (-), underscores (_), and slashes (/) are allowed. The value cannot start with a slash (/).

**Table 13** Health
Parameter	Type	Description
check_method	String	Definition Health check method, which can be HTTP or EXEC (commands). Range HTTP: Hypertext Transfer Protocol EXEC: commands check_method is not specified, the default value is check_method. During update, if check_method is not specified, an error will be reported.
cmd	String	Definition Command line mandatory when the health check method is EXEC. Range N/A
url	String	Definition Request address mandatory when the health check mode is HTTP. Range N/A
protocol	String	Definition Connection protocol. The default value is HTTP. Range HTTPS: Hypertext Transfer Protocol Secure HTTP: Hypertext Transfer Protocol WSS: WebSocket Secure WS: WebSocket
initial_delay_seconds	Integer	Definition Waiting time before the first detection. The default value is 30 and the minimum value is 1. Range N/A 30
timeout_seconds	Integer	Definition Timeout interval of the detection. The default value is 30 seconds, and the minimum value is 1 second. Range The minimum value is 1 second.
period_seconds	Integer	Definition Interval at which the health check is performed. The default value is 10 seconds, and the minimum value is 1 second. Range The minimum value is 1 second.
failure_threshold	Integer	Definition The minimum number of consecutive detection failures required after a successful detection. Detection is considered failed only when this threshold is reached. The default value is 3. The minimum value is 1. Range The minimum value is 1. 3

**Table 14** Affinity
Parameter	Type	Description
affinity_type	String	Definition Node affinity type. Range AFFINITY: affinity ANTI_AFFINITY: anti-affinity
required	Boolean	Definition Specifies whether to set strong affinity. Range true: Strong affinity is set. false: Strong affinity is not set.
selection_mode	String	Definition Node selection mode. Range IP
targets	Map<String,String>	Definition List selected using the preceding method, whose length cannot exceed 20.

**Table 15** ServiceSecurityConfig
Parameter	Type	Description
user_id	Long	Definition user_id set during service container startup. This parameter can be left empty by default. Range [1, 60000]
group_id	Long	Definition group_id set during service container startup. This parameter can be left empty by default. Range [1, 60000]

**Table 16** AdvancedConfig
Parameter	Type	Description
deploy_timeout_minutes	String	Definition Deployment timeout.
upgrade_config	UpgradeConfig object	Definition Rolling upgrade configuration.
service_secret	ServiceSecret object	Definition Service secret.
dynamic_routing_enable	Boolean	Definition Intelligent routing switch.
strategy	String	Definition Intelligent routing policy.
ems_enable	Boolean	Definition EMS acceleration switch.
metric_api_scheme	String	Definition Intelligent routing metric collection scheme.
metric_api_port	String	Definition Intelligent routing metric collection port.
metric_api_path	String	Definition Intelligent routing metric collection URL.
custom_metrics_path	String	Definition Custom metric collection URL.
port	Integer	Definition Container port.
protocol	String	Definition Container request protocol. When WSS or WS is selected, the service API is upgraded to WebSocket. WebSocket cannot be enabled together with Traffic Limit. Range HTTP: HTTP protocol HTTPS: HTTPS protocol WSS: WebSocket Secure protocol WS: WebSocket protocol TCP: Transmission Control Protocol NA: No protocol used

**Table 17** UpgradeConfig
Parameter	Type	Description
type	String	Definition Upgrade type. Range ROLLING: rolling upgrade (default value)
rolling_update	RollingUpdate object	Definition Rolling update configuration.

**Table 18** RollingUpdate
Parameter	Type	Description
max_surge	String	Definition Maximum number of pods that can be started during rolling update. Range 1/ %to 100% 1%.
max_unavailable	String	Definition Maximum number of pods that can be deleted during rolling update. Range 1/ %to 100% 1%.

**Table 19** ServiceSecret
Parameter	Type	Description
secret_enable	Boolean	Definition Specifies whether to enable the key. Range true: The secret is enabled. false: The secret is disabled. false
secret_type	String	Definition Secret type. Range custom: custom secret. [dew: DEW secret.] (tag:hws,hws_hk,fcs)
secret_volumes	Array of SecretVolume objects	Definition Secret volume mounting.
group_enable	Boolean	Definition Specifies whether to enable the image user group. Range true: The user group of the image is enabled. false: The user group of the image is not enabled. false
group_id	Long	Definition User group ID of the image. Range 1000 to 4294967294

**Table 20** SecretVolume
Parameter	Type	Description
secret_name	String	Definition Secret name. Range N/A
secret_key	String	Definition Secret key. Range N/A
secret_value	String	Definition Secret value. Range N/A
mount_path	String	Definition Mount path. Range The value must start and end with a slash (/) and can contain at most 255 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.

**Table 21** FrozenInfo
Parameter	Type	Description
scene	String	Freezing scenario. Options: ARREAR: frozen due to arrears POLICE: frozen due to public security issues ILLEGAL: frozen due to abuse
effect	Integer	Impacts of freezing. Options: 1: can be released after being frozen 2: cannot be released after being frozen

Example Requests

Modify the number of instances.

PUT https://{endpoint}/v2/{project_id}/services/{service_id}/deployments/{deployment_name}/scale

{
  "count" : 1
}

Example Responses

Status code: 200

The service scaling is successful.

{
  "deployment_name" : "instance-35cb",
  "status" : "running",
  "count" : 2
}

Status Codes

Status Code	Description
200	The service scaling is successful.

Error Codes

See Error Codes.

Parent topic: Service Deployment Instance Management

Previous topic: Service Deployment Instance Management

Next topic: Obtaining the Service Deployment Instance List