如何在Terraform的map(any)变量中配置list(string)类型node_taints

问题根源在于：使用map(any)类型时，Terraform会根据默认值自动推断map内每个元素的类型。你的默认值里vm1的node_taints是null，vm2的是list(string)，两种类型不兼容，导致Terraform无法统一推断，从而抛出错误。

下面提供两种解决方案：

方案一：统一node_taints类型 #

将vm1中的node_taints从null改为空列表[]，这样两个元素的node_taints都属于list(string)类型，类型统一后即可解决报错。修改后的默认值片段：

vm1 = {
  user_agents_name         = "upool01"
  user_agents_size         = "Standard_D2s_v5"
  user_agents_count        = "1"
  user_agents_os_disk_size = "32"
  max_pods                 = "20"
  orchestrator_version     = "1.27.7"
  node_taints              = []  # 将null改为空列表
  node_labels = {
    "commponent" = "idx"
    "keyexample" = "valueexample"
  }
}

方案二：显式定义结构化类型（推荐） #

用map(object({...}))替代map(any)，显式声明每个字段的类型，并将node_taints设为可选的list(string)类型，允许其值为null。这种方式比map(any)更安全，能提前发现类型错误：

variable "usernodepool_vm" {
  description = "VM of AKS Cluster"
  type = map(object({
    user_agents_name         = string
    user_agents_size         = string
    user_agents_count        = string
    user_agents_os_disk_size = string
    max_pods                 = string
    orchestrator_version     = string
    node_taints              = optional(list(string), null)
    node_labels              = map(string)
  }))
  default = {
    vm1 = {
      user_agents_name         = "upool01"
      user_agents_size         = "Standard_D2s_v5"
      user_agents_count        = "1"
      user_agents_os_disk_size = "32"
      max_pods                 = "20"
      orchestrator_version     = "1.27.7"
      node_taints              = null
      node_labels = {
        "commponent" = "idx"
        "keyexample" = "valueexample"
      }
    }
    vm2 = {
      user_agents_name         = "upool02"
      user_agents_size         = "Standard_D2s_v5"
      user_agents_count        = "1"
      user_agents_os_disk_size = "32"
      max_pods                 = "20"
      orchestrator_version     = "1.27.7"
      node_taints              = ["key01=value01:NoSchedule"]
      node_labels = {
        "compponent" = "oom"
        "no" = "numeric"
      }
    }
  }
}

内容的提问来源于stack exchange，提问作者gotothesky