Manage HPA elastic policies with the saectl tool-Serverless App Engine(SAE)-阿里云帮助中心

SAE elastic policies map to the Kubernetes HorizontalPodAutoscaler (HPA) resource type. Use the saectl tool to create, view, update, and delete elastic policies for your applications.

Prerequisites

Before you begin, ensure that you have:

The saectl tool installed, with an AccessKey ID, AccessKey secret, and application deployment region configured. For setup instructions, see Install and configure the saectl tool

Manage elastic policies

Create an elastic policy

Create a file named hpa.yaml with the following configuration. For a description of each field, see K8s YAML configuration reference.

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  annotations:
    sae.aliyun.com/alias: test-app        # Optional: nickname for the elastic policy
    sae.aliyun.com/enabled: "true"         # Optional: enable or disable the policy
  name: metric-test-app                    # Fixed format: metric-${application_name}
  namespace: default
spec:
  behavior:                                # Optional: scaling behavior
    scaleDown:
      policies:
      - periodSeconds: 15                  # Fixed by SAE, do not modify
        type: Pods                         # Fixed by SAE, do not modify
        value: 10                          # Scale-in step size
      stabilizationWindowSeconds: 200
    scaleUp:
      policies:
      - periodSeconds: 15                  # Fixed by SAE, do not modify
        type: Pods                         # Fixed by SAE, do not modify
        value: 5                           # Scale-out step size
      stabilizationWindowSeconds: 100
  maxReplicas: 8
  metrics:
  - resource:
      name: cpu
      target:
        averageUtilization: 80
        type: Utilization
    type: Resource
  - resource:
      name: memory
      target:
        averageUtilization: 80
        type: Utilization
    type: Resource
  - pods:
      metric:
        name: tcpActiveConn
      target:
        averageValue: "100"
        type: AverageValue
    type: Pods
  - pods:
      metric:
        name: tcpTotalConn
      target:
        averageValue: "100"
        type: AverageValue
    type: Pods
  - external:
      metric:
        name: qps
      target:
        averageValue: "100"
        type: AverageValue
    type: External
  - external:
      metric:
        name: rt
      target:
        averageValue: 100k
        type: AverageValue
    type: External
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: test-app                         # Fixed format: ${application_name}

From the directory where hpa.yaml is located, run:
```
saectl apply -f hpa.yaml
```

List elastic policies

saectl get hpa -l sae.aliyun.com/app-name=<application-name> -n <namespace>

Parameter	Description
`-l sae.aliyun.com/app-name=<application-name>`	Filter by application name. If omitted, returns all HPAs in scope.
`-n <namespace>`	Namespace ID. If omitted, the default namespace is used.

The output lists the following fields:

Field	Description
`NAME`	Elastic policy name. Format: `metric-${application_name}`.
`REFERENCEKIND`	Associated resource type.
`REFERENCENAME`	Name of the associated application.
`MINPODS`	Minimum number of instances.
`MAXPODS`	Maximum number of instances.
`CURRENTREPLICAS`	Current number of running instances.
`DESIREDREPLICAS`	Target number of instances.
`ENABLED`	Whether the elastic policy is enabled.

View elastic policy details

You can view the details of an elastic policy using the get or describe command. The output contains the configuration items of the elastic policy.

Get structured output

saectl get hpa <hpa-name> -o {yaml | json} -n <namespace>

Parameter	Description
`<hpa-name>`	Name of the HPA. Format: `metric-${application_name}`.
`-o`	Output format: `yaml` or `json`.
`-n <namespace>`	Namespace ID. If omitted, the default namespace is used.

Describe an elastic policy

saectl describe hpa <hpa-name> -n <namespace>

Parameter	Description
`<hpa-name>`	Name of the HPA. Format: `metric-${application_name}`.
`-n <namespace>`	Namespace ID. If omitted, the default namespace is used.

Update an elastic policy

Two methods are available: edit the policy online with edit, or apply an updated YAML file with apply.

Edit online

Open the policy configuration in your editor:
```
saectl edit hpa <hpa-name> -n <namespace>
```
Modify the configuration as needed, then save and close the file for the changes to take effect.

Apply from a YAML file

Modify the existing hpa.yaml file, then run:

saectl apply -f hpa.yaml

Delete an elastic policy

saectl delete hpa <hpa-name> -n <namespace>

Parameter	Description
`<hpa-name>`	Name of the HPA. Format: `metric-${application_name}`.
`-n <namespace>`	Namespace ID. If omitted, the default namespace is used.

K8s YAML configuration reference

SAE elastic policies use the Kubernetes HPA resource type (autoscaling/v2). The following table describes the supported configuration fields.

Field	Description	Required for creation	Updatable	Constraints
`metadata.name`	Elastic policy name. Use the fixed format `metric-${application_name}`.	Yes	No	—
`metadata.namespace`	Namespace where the elastic policy resides.	—	No	—
`metadata.annotations: sae.aliyun.com/alias`	Nickname for the elastic policy.	—	No	—
`metadata.annotations: sae.aliyun.com/enabled`	Whether to enable the elastic policy.	—	Yes	Valid values: `"true"` (enabled), `"false"` (disabled)
`metadata.uid`	Elastic policy UID. Use the fixed format `metric-${application_ID}`.	—	No	—
`metadata.labels`	Label for the elastic policy. Use the fixed format `sae.aliyun.com/app-name: ${application_name}`.	—	No	—
`spec.behavior.scaleDown.policies[i]`	Scale-in step size configuration. Set `value` to configure the step size. `periodSeconds` (15) and `type` (Pods) are fixed by SAE and cannot be modified.	Optional	—	`periodSeconds` and `type` are fixed
`spec.behavior.scaleUp.policies[i]`	Scale-out step size configuration. Set `value` to configure the step size. `periodSeconds` (15) and `type` (Pods) are fixed by SAE and cannot be modified.	Optional	—	`periodSeconds` and `type` are fixed
`spec.maxReplicas`	Maximum number of instances.	—	—	—
`spec.minReplicas`	Minimum number of instances.	—	—	—
`spec.scaleTargetRef`	Target application reference. Required for creation, not updatable.	Yes	No	Fixed format: `apiVersion: apps/v1`, `kind: Deployment`, `name: <application_name>`