如何用Terraform检查Azure APIM自动缩放设置并解决重复报错

问题根源 #

Azure规定每个API Management资源只能关联一个自动缩放设置。首次部署成功后，再次运行Terraform计划时，原代码使用的azapi_resource尝试创建同名/同目标的自动缩放设置，触发SettingAlreadyExists（409冲突）错误，核心原因是Terraform未正确匹配已存在的远程资源状态，或资源定义未遵循Azure的唯一性限制。

解决方案 #

方案1：使用AzureRM官方自动缩放资源（推荐） #

AzureRM提供了专门的azurerm_autoscale_setting资源类型，能自动处理目标资源的唯一性关联，避免重复创建冲突。替换原有azapi_resource定义为以下代码：

resource "azurerm_autoscale_setting" "apim_scaling" {
  name                = "Auto-scale-capacity-70-percentage"
  resource_group_name = data.azurerm_resource_group.rg.name
  location            = var.resource_group.location
  target_resource_id  = azurerm_api_management.apim[0].id

  enabled = true

  notification {
    email {
      custom_emails = ["cloud@test.com"]
      send_to_subscription_administrator    = false
      send_to_subscription_co_administrators = false
    }
    operation = "Scale"
  }

  profile {
    name = "Auto created default scale condition"

    capacity {
      default = 1
      maximum = 3
      minimum = 1
    }

    rule {
      metric_trigger {
        metric_name        = "Capacity"
        metric_namespace   = "microsoft.apimanagement/service"
        metric_resource_id = azurerm_api_management.apim[0].id
        operator           = "GreaterThan"
        statistic          = "Average"
        threshold          = 70
        time_aggregation   = "Average"
        time_grain         = "PT1M"
        time_window        = "PT10M"
      }

      scale_action {
        direction = "Increase"
        type      = "ChangeCount"
        value     = 1
        cooldown  = "PT5M"
      }
    }

    rule {
      metric_trigger {
        metric_name        = "Capacity"
        metric_namespace   = "microsoft.apimanagement/service"
        metric_resource_id = azurerm_api_management.apim[0].id
        operator           = "LessThan"
        statistic          = "Average"
        threshold          = 65
        time_aggregation   = "Average"
        time_grain         = "PT1M"
        time_window        = "PT10M"
      }

      scale_action {
        direction = "Decrease"
        type      = "ChangeCount"
        value     = 1
        cooldown  = "PT5M"
      }
    }
  }

  tags = local.tags

  depends_on = [azurerm_api_management.apim]
}

方案2：保留azapi_resource并添加前置检查 #

若需继续使用azapi_resource，可通过Terraform数据源查询目标APIM是否已关联自动缩放设置，再通过count控制资源是否创建：

步骤1：添加数据源查询已有自动缩放设置

# 查询资源组下所有自动缩放设置
data "azurerm_autoscale_settings" "apim_existing" {
  resource_group_name = data.azurerm_resource_group.rg.name
}

# 判断目标APIM是否已有关联的自动缩放设置
locals {
  apim_has_autoscale = anytrue([
    for setting in data.azurerm_autoscale_settings.apim_existing.settings :
    setting.target_resource_id == azurerm_api_management.apim[0].id
  ])
}

步骤2：修改azapi_resource定义，通过count控制创建

resource "azapi_resource" "apim_scaling" {
  # 仅当无已存在的自动缩放设置时才创建
  count               = local.apim_has_autoscale ? 0 : 1
  type                = "Microsoft.Insights/autoscalesettings@2022-10-01"
  name                = "Auto-scale-capacity-70-percentage"
  location            = var.resource_group.location
  parent_id           = data.azurerm_resource_group.rg.id
  tags                = local.tags
  body                = jsonencode({
    properties = {
      enabled = true
      name    = "Auto-scale-capacity-70-percentage"
      notifications = [
        {
          email = {
            customEmails = [
              "cloud@test.com"
            ]
            sendToSubscriptionAdministrator    = false
            sendToSubscriptionCoAdministrators = false
          }
          operation = "Scale"
        }
      ]
      profiles = [
        {
          capacity = {
            default = "1"
            maximum = "3"
            minimum = "1"
          }
          name = "Auto created default scale condition"
          rules = [
            {
              metricTrigger = {
                dimensions             = []
                dividePerInstance      = false
                metricName             = "Capacity"
                metricNamespace        = "microsoft.apimanagement/service"
                metricResourceLocation = azurerm_api_management.apim[0].location
                metricResourceUri      = azurerm_api_management.apim[0].id
                operator               = "GreaterThan"
                statistic              = "Average"
                threshold              = 70
                timeAggregation        = "Average"
                timeGrain              = "PT1M"
                timeWindow             = "PT10M"
              }
              scaleAction = {
                cooldown  = "PT5M"
                direction = "Increase"
                type      = "ChangeCount"
                value     = "1"
              }
            },
            {
              metricTrigger = {
                dimensions             = []
                dividePerInstance      = false
                metricName             = "Capacity"
                metricNamespace        = "microsoft.apimanagement/service"
                metricResourceLocation = azurerm_api_management.apim[0].location
                metricResourceUri      = azurerm_api_management.apim[0].id
                operator               = "LessThan"
                statistic              = "Average"
                threshold              = 65
                timeAggregation        = "Average"
                timeGrain              = "PT1M"
                timeWindow             = "PT10M"
              }
              scaleAction = {
                cooldown  = "PT5M"
                direction = "Decrease"
                type      = "ChangeCount"
                value     = "1"
              }
            }
          ]
        }
      ]
      targetResourceLocation = azurerm_api_management.apim[0].location
      targetResourceUri      = azurerm_api_management.apim[0].id
    }
  })
  depends_on = [
    azurerm_api_management.apim
  ]
}

关键说明 #

• 方案1优先推荐：官方资源类型会自动处理Azure的唯一性约束，无需额外编写检查逻辑，且Terraform状态匹配更可靠。
• 方案2适用于必须使用azapi_resource的场景：通过数据源查询+条件创建，避免重复提交创建请求。

内容的提问来源于stack exchange，提问作者Madhu