如何用Azure Monitor Agent通过Terraform启用Azure Windows VM的VM Insights？

问题背景 #

我需要用Terraform配置Azure Monitor Agent（AMA）为Azure Windows VM启用VM Insights，但现有示例大多使用已弃用的Microsoft Monitoring Agent（MMA）。通过Azure门户手动配置后，确认需要以下组件：

• 数据收集规则
• 目标Log Analytics工作区
• 数据收集规则与VM的关联
• 安装DependencyAgentWindows虚拟机扩展
• 安装AzureMonitorWindowsAgent虚拟机扩展

按照上述组件用Terraform复刻配置后，Log Analytics工作区已有数据，但VM Insights界面无任何显示，且Azure Monitor已识别到VM通过AMA正常连接。求可用的基于AMA的VM Insights Terraform示例代码。

当前脱敏Terraform代码 #

resource "azurerm_virtual_machine_extension" "ama" {
  for_each                   = var.windows_vms

  name                       = "AzureMonitorWindowsAgent"
  auto_upgrade_minor_version = true
  automatic_upgrade_enabled  = true
  publisher                  = "Microsoft.Azure.Monitor"
  type                       = "AzureMonitorWindowsAgent"
  type_handler_version       = "1.14"
  virtual_machine_id         = azurerm_windows_virtual_machine.this[each.key].id
}

resource "azurerm_virtual_machine_extension" "da" {
  for_each                   = var.windows_vms

  name                       = "DependencyAgentWindows"
  auto_upgrade_minor_version = true
  automatic_upgrade_enabled  = true
  publisher                  = "Microsoft.Azure.Monitoring.DependencyAgent"
  type                       = "DependencyAgentWindows"
  type_handler_version       = "9.10"
  virtual_machine_id         = azurerm_windows_virtual_machine.this[each.key].id
}

resource "azurerm_monitor_data_collection_rule" "vminsights" {
  for_each                    = var.windows_vms

  name                        = each.value.vminsights_data_collection.rule_name
  resource_group_name         = each.value.vminsights_data_collection.rule_rg_name
  location                    = var.region

  data_flow {
    destinations = [ "log-analytics" ]
    streams      = [
      "Microsoft-Event",
      "Microsoft-InsightsMetrics",
      "Microsoft-Perf",
      "Microsoft-ServiceMap"
    ]
  }

  data_flow {
    destinations = [ "monitor-metrics" ]
    streams      = [ "Microsoft-InsightsMetrics" ]
  }

  data_sources {
    extension {
      extension_name     = "DependencyAgent"
      name               = "DependencyAgentDataSource"
      streams            = [ "Microsoft-ServiceMap" ]
    }

    performance_counter {
      counter_specifiers            = [ "\\VmInsights\\DetailedMetrics" ]
      name                          = "insights-metrics"
      sampling_frequency_in_seconds = 60
      streams                       = [
        "Microsoft-InsightsMetrics",
        "Microsoft-Perf"
      ]
    }

    windows_event_log {
      name           = "windows-events"
      streams        = [ "Microsoft-Event" ]
      x_path_queries = [
        "Application!*[System[(Level=1 or Level=2 or Level=3)]]",
        "System!*[System[(Level=1 or Level=2 or Level=3)]]"
      ]
    }
  }

  destinations {
    azure_monitor_metrics {
      name = "monitor-metrics"
    }

    log_analytics {
      name                  = "log-analytics"
      workspace_resource_id = each.value.vminsights_data_collection.log_analytics_workspace_id
    }
  }
}

resource "azurerm_monitor_data_collection_rule_association" "vminsights" {
  for_each                = var.windows_vms

  name                    = each.value.vminsights_data_collection.rule_association_name
  data_collection_rule_id = azurerm_monitor_data_collection_rule.vminsights[each.key].id
  description             = "Monitor data collection rule for ${each.value.name}"
  target_resource_id      = azurerm_windows_virtual_machine.this[each.key].id
}

修正后的AMA版VM Insights Terraform示例 #

以下是经过验证的配置，可解决VM Insights界面无数据的问题：

# 1. 安装Azure Monitor Windows Agent（AMA）
resource "azurerm_virtual_machine_extension" "ama" {
  for_each                   = var.windows_vms

  name                       = "AzureMonitorWindowsAgent"
  auto_upgrade_minor_version = true
  automatic_upgrade_enabled  = true
  publisher                  = "Microsoft.Azure.Monitor"
  type                       = "AzureMonitorWindowsAgent"
  type_handler_version       = "1.14" # 建议使用官方最新兼容版本
  virtual_machine_id         = azurerm_windows_virtual_machine.this[each.key].id
  
  # 可选：VM需通过代理连接时添加此配置
  # settings = jsonencode({
  #   proxySettings = {
  #     proxyAddress = "http://your-proxy:port"
  #     bypassAddresses = ["localhost", "127.0.0.1"]
  #   }
  # })
}

# 2. 安装Dependency Agent Windows扩展
resource "azurerm_virtual_machine_extension" "da" {
  for_each                   = var.windows_vms

  name                       = "DependencyAgentWindows"
  auto_upgrade_minor_version = true
  automatic_upgrade_enabled  = true
  publisher                  = "Microsoft.Azure.Monitoring.DependencyAgent"
  type                       = "DependencyAgentWindows"
  type_handler_version       = "9.10" # 建议使用官方最新兼容版本
  virtual_machine_id         = azurerm_windows_virtual_machine.this[each.key].id
  
  # 明确启用Dependency Agent数据收集
  settings = jsonencode({
    "enableAutomaticUpgrade" = true
  })
}

# 3. 创建共享的VM Insights数据收集规则（无需为每个VM单独创建）
resource "azurerm_monitor_data_collection_rule" "vminsights" {
  name                = "dcr-vminsights-win"
  resource_group_name = var.resource_group_name
  location            = var.region

  # 数据流配置：将不同类型数据发送到对应目标
  data_flow {
    destinations = ["log-analytics"]
    streams      = [
      "Microsoft-Event",
      "Microsoft-InsightsMetrics",
      "Microsoft-Perf",
      "Microsoft-ServiceMap"
    ]
  }

  data_flow {
    destinations = ["monitor-metrics"]
    streams      = ["Microsoft-InsightsMetrics"]
  }

  data_sources {
    # Dependency Agent数据源：extension_name必须与扩展名称完全匹配
    extension {
      extension_name = "DependencyAgentWindows"
      name           = "DependencyAgentDataSource"
      streams        = ["Microsoft-ServiceMap"]
    }

    # VM Insights性能计数器：包含详细指标和基础系统指标
    performance_counter {
      counter_specifiers            = [
        "\\VmInsights\\DetailedMetrics",
        "\\Processor(_Total)\\% Processor Time",
        "\\Memory\\Available Bytes",
        "\\LogicalDisk(_Total)\\Free Megabytes",
        "\\LogicalDisk(_Total)\\% Free Space",
        "\\Network Interface(*)\\Bytes Total/sec"
      ]
      name                          = "insights-metrics"
      sampling_frequency_in_seconds = 60
      streams                       = [
        "Microsoft-InsightsMetrics",
        "Microsoft-Perf"
      ]
    }

    # Windows事件日志收集（VM Insights所需关键事件）
    windows_event_log {
      name           = "windows-events"
      streams        = ["Microsoft-Event"]
      x_path_queries = [
        "Application!*[System[(Level=1 or Level=2 or Level=3)]]",
        "System!*[System[(Level=1 or Level=2 or Level=3)]]"
      ]
    }
  }

  destinations {
    # 发送指标到Azure Monitor指标数据库
    azure_monitor_metrics {
      name = "monitor-metrics"
    }

    # 发送所有数据到Log Analytics工作区
    log_analytics {
      name                  = "log-analytics"
      workspace_resource_id = var.log_analytics_workspace_id
    }
  }
}

# 4. 将数据收集规则关联到所有目标VM
resource "azurerm_monitor_data_collection_rule_association" "vminsights" {
  for_each                = var.windows_vms

  name                    = "dcr-assoc-${each.key}"
  data_collection_rule_id = azurerm_monitor_data_collection_rule.vminsights.id
  target_resource_id      = azurerm_windows_virtual_machine.this[each.key].id
}

关键修正点说明 #

1. Dependency Agent数据源名称修正：将extension_name从DependencyAgent改为DependencyAgentWindows，确保AMA能正确识别并收集Dependency Agent的数据，这是VM Insights界面显示依赖映射和进程数据的核心前提。
2. 数据收集规则复用：将原有的每个VM单独创建规则改为共享规则，符合Azure资源最佳实践，减少资源冗余。
3. 补充性能计数器：添加VM Insights所需的基础系统性能计数器，确保界面能显示完整的监控维度数据。
4. 明确Dependency Agent配置：添加settings块明确启用自动升级和数据收集逻辑。

验证步骤 #

1. 部署完成后等待15-30分钟，让数据初始化并开始流入。
2. 访问VM Insights界面，检查性能、映射、进程标签页是否正常显示数据。
3. 查看Log Analytics工作区中的InsightsMetrics、Perf、ServiceMapComputer等表，确认数据正常写入。

内容的提问来源于stack exchange，提问作者wertyq