DocsICP FilingConsole
官方文档
首页

ACS::AlarmTrigger

更新时间：
复制 MD 格式
产品详情
我的收藏

The ACS::AlarmTrigger action monitors a CloudMonitor metric and triggers subsequent tasks in an OOS template when the metric crosses a threshold. Use it to build self-healing workflows -- for example, automatically rebooting an ECS instance when its CPU usage exceeds 90%.

How it works

  1. An OOS execution containing ACS::AlarmTrigger starts in Pending status.

  2. When the specified metric reaches the threshold, the execution transitions to Active status.

  3. OOS runs the subsequent tasks defined in the template to remediate the alert.

Important

AlarmTrigger supports two types of CloudMonitor metrics:

  • ECS native metrics -- available without additional setup.

  • Agent-collected metrics -- require the CloudMonitor agent on the instance. To install the agent, open the CloudMonitor console, go to Host Monitoring, select the target instance, and click Install. See the CloudMonitor metric descriptions to determine which type a metric belongs to.

Limitations

  • A template can contain only one trigger action.

  • The trigger action must be the first task in the template.

  • Trigger actions cannot be used in child templates (embedded templates).

Properties

Property Required Type Description
Namespace Yes String CloudMonitor namespace of the service to monitor (for example, acs_ecs_dashboard for ECS). Query valid values with the DescribeMetricMetaList operation.
MetricName Yes String Metric name (for example, cpu_total). Query valid values with DescribeMetricMetaList.
Statistics Yes String Aggregation method applied to the metric data, such as Average, Maximum, or Minimum. Query valid values with DescribeMetricMetaList.
ComparisonOperator Yes String Operator that compares the metric value against the threshold. See Comparison operators.
Threshold Yes String Value that triggers the alert (for example, "90" for 90% CPU usage).
Resources Yes String JSON array specifying the resources to monitor. See Resource formats.
Times Yes Integer Number of consecutive breaches required before the alert fires.
Interval No Integer Evaluation interval in seconds. Default: 60.
SilenceTime No Integer Mute duration in seconds. During this period, only one alert notification is sent even if the metric continuously exceeds the threshold. Default: 86400 (24 hours). Minimum: 3600 (1 hour).
Period No Integer Aggregation period in seconds. Default: the metric's original reporting period (typically 60 for ECS metrics).
Tags No JSON Filter monitored resources by tag. Format: [{"Key": "k1", "Value": "v1"}]. Mutually exclusive with ResourceGroup.
ResourceGroup No String Filter monitored resources by resource group ID (for example, rg-xxxx). Mutually exclusive with Tags.

Comparison operators

Value Meaning
GreaterThanOrEqualToThreshold >= threshold
GreaterThanThreshold > threshold
LessThanOrEqualToThreshold <= threshold
LessThanThreshold < threshold
NotEqualToThreshold != threshold
GreaterThanYesterday Increase from the same time yesterday
LessThanYesterday Decrease from the same time yesterday
GreaterThanLastWeek Increase from the same time last week
LessThanLastWeek Decrease from the same time last week
GreaterThanLastPeriod Increase from the previous period
LessThanLastPeriod Decrease from the previous period

Resource formats

Scope Format
All resources under the current account [{"resource":"_ALL"}]
A specific instance [{"instanceId":"i-bp123467zxcvb"}]
A specific disk partition [{"instanceId":"i-bp123467zxcvb","device":"/dev/vda1"}]
Multiple disk partitions [{"instanceId":"i-bp123467zxcvb","device":"/dev/vda1"},{"instanceId":"i-bp123467zxcvb","device":"/dev/vdb1"}]

Outputs

Use the Outputs block to extract values from the alert event JSON. Specify a ValueSelector using a JSON path expression (for example, .instanceId).

Alert event JSON structure:

{
  "curLevel": "INFO",
  "Minimum": "34.00",
  "Maximum": "95.00",
  "instanceId": "i-abc12345zxcv",
  "Average": "85.00",
  "ruleName": "alarmtrigger-13012345678-exec-2130c0c073fa487098d3",
  "userId": "13012345678",
  "timestamp": "1598349720000",
  "executionId": "exec-2130c0c073fa487098d3",
  "sourceAliUid": "13012345678"
}
Field Description
instanceId ID of the instance that triggered the alert
curLevel Alert severity level
Average / Minimum / Maximum Aggregated metric values
ruleName Name of the alert rule
userId Alibaba Cloud account ID
timestamp Unix timestamp (milliseconds) of the alert event
executionId OOS execution ID

Syntax

YAML

Tasks:
  - Name: <task-name>
    Action: 'ACS::AlarmTrigger'
    Properties:
      Namespace: 'acs_ecs_dashboard'
      MetricName: 'cpu_total'
      Statistics: 'Average'
      ComparisonOperator: 'GreaterThanThreshold'
      Threshold: '90'
      Resources: '[{"resource":"_ALL"}]'
      Times: 1
      Interval: 60
      SilenceTime: 3600
    Outputs:
      instanceId:
        Type: String
        ValueSelector: .instanceId

JSON

{
  "Tasks": [
    {
      "Name": "<task-name>",
      "Action": "ACS::AlarmTrigger",
      "Properties": {
        "Namespace": "acs_ecs_dashboard",
        "MetricName": "cpu_total",
        "Statistics": "Average",
        "ComparisonOperator": "GreaterThanThreshold",
        "Threshold": "90",
        "Resources": "[{\"resource\":\"_ALL\"}]",
        "Times": 1,
        "Interval": 60,
        "SilenceTime": 3600
      },
      "Outputs": {
        "instanceId": {
          "Type": "String",
          "ValueSelector": ".instanceId"
        }
      }
    }
  ]
}

Example: Reboot an ECS instance on high CPU

This template monitors ECS instances matching specific tags. When an instance's average CPU usage exceeds the threshold in a 1-minute evaluation window, OOS verifies the instance has the expected tag, then reboots it.

Prerequisites:

  • The CloudMonitor agent is installed on the target instances

  • The OOSServiceRole RAM role exists (or specify a custom role)

YAML

FormatVersion: OOS-2019-06-01
Description:
  en: Reboot ECS instance with specified tag when its CPU utilization exceeded threshold. The selected instance must already have the Cloud Monitor agent installed.
  name-en: ACS-ECS-RebootInstanceAtHighCpuByTags
  categories:
    - alarm-trigger
Parameters:
  tags:
    Type: Json
    Description:
      en: The tags to select ECS instances.
    AssociationProperty: Tags
  threshold:
    Type: Number
    Description:
      en: The CPU utilization threshold.
  silenceTime:
    Type: Number
    Description:
      en: The silence time of alarm (seconds).
    Default: 60
  OOSAssumeRole:
    Description:
      en: The RAM role to be assumed by OOS.
    Type: String
    Default: OOSServiceRole
RamRole: '{{ OOSAssumeRole }}'
Tasks:
  - Name: alarmTrigger
    Action: 'ACS::AlarmTrigger'
    Description:
      en: Set the CPU utilization alarm for ECS instance.
    Properties:
      Namespace: acs_ecs_dashboard
      MetricName: cpu_total
      Statistics: Average
      ComparisonOperator: GreaterThanThreshold
      Threshold: '{{threshold}}'
      Times: 1
      SilenceTime: '{{ silenceTime }}'
      Period: 60
      Interval: 60
    Outputs:
      InstanceId:
        Type: String
        ValueSelector: .instanceId
  - Name: CheckForInstances
    Action: 'ACS::CheckFor'
    Description:
      en: Check ECS instance has specified tag.
    OnError: 'ACS::END'
    Properties:
      Service: ECS
      API: DescribeInstances
      Parameters:
        Tags: '{{ tags }}'
        InstanceIds: '["{{ alarmTrigger.instanceId }}"]'
      PropertySelector: TotalCount
      DesiredValues:
        - 1
  - Name: RebootInstance
    Action: 'ACS::ECS::RebootInstance'
    Description:
      en: Restarts the ECS instances.
    Properties:
      instanceId: '{{ alarmTrigger.instanceId }}'

JSON

{
  "FormatVersion": "OOS-2019-06-01",
  "Description": {
    "en": "Reboot ECS instance with specified tag when its CPU utilization exceeded threshold. The selected instance must already have the Cloud Monitor agent installed.",
    "name-en": "ACS-ECS-RebootInstanceAtHighCpuByTags",
    "categories": ["alarm-trigger"]
  },
  "Parameters": {
    "tags": {
      "Type": "Json",
      "Description": { "en": "The tags to select ECS instances." },
      "AssociationProperty": "Tags"
    },
    "threshold": {
      "Type": "Number",
      "Description": { "en": "The CPU utilization threshold." }
    },
    "silenceTime": {
      "Type": "Number",
      "Description": { "en": "The silence time of alarm (seconds)." },
      "Default": 60
    },
    "OOSAssumeRole": {
      "Description": { "en": "The RAM role to be assumed by OOS." },
      "Type": "String",
      "Default": "OOSServiceRole"
    }
  },
  "RamRole": "{{ OOSAssumeRole }}",
  "Tasks": [
    {
      "Name": "alarmTrigger",
      "Action": "ACS::AlarmTrigger",
      "Description": { "en": "Set the CPU utilization alarm for ECS instance." },
      "Properties": {
        "Namespace": "acs_ecs_dashboard",
        "MetricName": "cpu_total",
        "Statistics": "Average",
        "ComparisonOperator": "GreaterThanThreshold",
        "Threshold": "{{threshold}}",
        "Times": 1,
        "SilenceTime": "{{ silenceTime }}",
        "Period": 60,
        "Interval": 60
      },
      "Outputs": {
        "InstanceId": {
          "Type": "String",
          "ValueSelector": ".instanceId"
        }
      }
    },
    {
      "Name": "CheckForInstances",
      "Action": "ACS::CheckFor",
      "Description": { "en": "Check ECS instance has specified tag." },
      "OnError": "ACS::END",
      "Properties": {
        "Service": "ECS",
        "API": "DescribeInstances",
        "Parameters": {
          "Tags": "{{ tags }}",
          "InstanceIds": "[\"{{ alarmTrigger.instanceId }}\"]"
        },
        "PropertySelector": "TotalCount",
        "DesiredValues": [1]
      }
    },
    {
      "Name": "RebootInstance",
      "Action": "ACS::ECS::RebootInstance",
      "Description": { "en": "Restarts the ECS instances." },
      "Properties": {
        "instanceId": "{{ alarmTrigger.instanceId }}"
      }
    }
  ]
}
Previous：Time Trigger ACS::TimerTriggerNext：ACS::EventTrigger