为什么“terraform import”没有读取我的配置？

Question

我正在使用 Go 构建一个 Terraform 插件/提供程序。我的架构中有以下内容（链接）：

Importer: &schema.ResourceImporter{
  State: resourceKubernetesClusterNodePoolImport,
}

这就是函数（链接）：

func resourceKubernetesClusterNodePoolImport(d *schema.ResourceData, m interface{}) ([]*schema.ResourceData, error) {
    apiClient := m.(*civogo.Client)

    clusterID, nodePoolID, err := utils.ResourceCommonParseID(d.Id())
    if err != nil {
        return nil, err
    }

    log.Printf("[INFO] retriving the node pool %s", nodePoolID)
    resp, err := apiClient.GetKubernetesCluster(clusterID)
    if err != nil {
        if resp != nil {
            return nil, err
        }
    }

    d.Set("cluster_id", resp.ID)
    for _, v := range resp.Pools {
        if v.ID == nodePoolID {
            d.Set("num_target_nodes", v.Count)
            d.Set("target_nodes_size", v.Size)
        }
    }

    return []*schema.ResourceData{d}, nil
}

当我在上面的函数顶部添加这些调试行时，结果（

reg

和

ok

）为空并且

false

：

reg, ok := d.GetOk("region")
log.Printf("resourceKubernetesClusterNodePoolImport reg %+v\n", reg)
log.Println("resourceKubernetesClusterNodePoolImport ok", ok)
log.Println("resourceKubernetesClusterNodePoolImport d.Get()", d.Get("region"))
if !ok {
    return nil, fmt.Errorf("[ERR] Please provide a region in your import configuration")
}

痕迹：

$ echo $TF_LOG
INFO
$ tf import civo_kubernetes_node_pool.my_pool 1b272dc1-836c-4ea7-8f20-4569f20ddbdd:d0c22622-54b3-41c2-9eae-24591737d5ad
2021-08-12T11:22:30.552+0800 [INFO]  Terraform version: 1.0.3
2021-08-12T11:22:30.552+0800 [INFO]  Go runtime version: go1.16.4
2021-08-12T11:22:30.552+0800 [INFO]  CLI args: []string{"/usr/local/bin/terraform", "import", "civo_kubernetes_node_pool.my_pool", "1b272dc1-836c-4ea7-8f20-4569f20ddbdd:d0c22622-54b3-41c2-9eae-24591737d5ad"}
2021-08-12T11:22:30.552+0800 [INFO]  Loading CLI configuration from /Users/zulh/.terraformrc
2021-08-12T11:22:30.553+0800 [INFO]  CLI command args: []string{"import", "civo_kubernetes_node_pool.my_pool", "1b272dc1-836c-4ea7-8f20-4569f20ddbdd:d0c22622-54b3-41c2-9eae-24591737d5ad"}
2021-08-12T11:22:30.629+0800 [INFO]  Failed to read plugin lock file .terraform/plugins/darwin_amd64/lock.json: open .terraform/plugins/darwin_amd64/lock.json: no such file or directory
2021-08-12T11:22:30.631+0800 [INFO]  provider: configuring client automatic mTLS
2021-08-12T11:22:31.284+0800 [INFO]  provider.terraform-provider-civo_v99.0.0: configuring server automatic mTLS: timestamp=2021-08-12T11:22:31.283+0800
2021-08-12T11:22:31.358+0800 [INFO]  provider: configuring client automatic mTLS
2021-08-12T11:22:31.404+0800 [INFO]  provider.terraform-provider-civo_v99.0.0: configuring server automatic mTLS: timestamp=2021-08-12T11:22:31.404+0800
2021-08-12T11:22:31.478+0800 [WARN]  ValidateProviderConfig from "provider[\"registry.terraform.io/civo/civo\"]" changed the config value, but that value is unused
civo_kubernetes_node_pool.my_pool: Importing from ID "1b272dc1-836c-4ea7-8f20-4569f20ddbdd:d0c22622-54b3-41c2-9eae-24591737d5ad"...
2021-08-12T11:22:31.479+0800 [INFO]  provider.terraform-provider-civo_v99.0.0: 2021/08/12 11:22:31 resourceKubernetesClusterNodePoolImport reg: timestamp=2021-08-12T11:22:31.479+0800      <-- this line
2021-08-12T11:22:31.479+0800 [INFO]  provider.terraform-provider-civo_v99.0.0: 2021/08/12 11:22:31 resourceKubernetesClusterNodePoolImport ok false: timestamp=2021-08-12T11:22:31.479+0800 <-- this line
2021-08-12T11:22:31.479+0800 [INFO]  provider.terraform-provider-civo_v99.0.0: 2021/08/12 11:22:31 resourceKubernetesClusterNodePoolImport d.Get(): timestamp=2021-08-12T11:22:31.479+0800  <-- this line
╷
│ Error: [ERR] Please provide a region in your import configuration
│
│
╵

问题是，在运行上面的

region

命令之前，我已在 Terraform 配置 main.tf 文件中添加了

tf import

（并保存）。这是我的 main.tf 文件：

resource "civo_kubernetes_cluster" "my_cluster" {
    name = "my-cluster"
}

resource "civo_kubernetes_node_pool" "my_pool" {
    region = "LON1" <-- this line
}

为什么我的 Terraform 插件/提供程序将该区域视为空？

根据我的理解，Terraform 在运行导入逻辑之前会首先读取配置文件。我错了吗？我在这里错过了什么吗？

---

为了完整起见，我使用 Terraform v1.0.3，这是我的 provider.tf 文件...

terraform {
  required_providers {
    civo = {
      source  = "civo/civo"
      version = "99.0.0"
    }
  }
}

provider "civo" {
    token = “my-api-key-here”
}

...其中

99.0.0

是我仅为测试而构建的版本。当我发布此问题时，该提供商的最新/产品版本是 0.10.9。

Answer 1

Importer.State

函数无法访问现有配置，它通过

id

提供空资源状态 - 参见代码。

作为解决方法，您可以允许传递

region

通过下划线（或任何不与 ids 或区域名称中的符号冲突的符号）导入命令，如下所示：

terraform import civo_kubernetes_node_pool.my_pool <poolid>_<region>

并在导入器函数中处理这个问题：

func resourceKubernetesClusterNodePoolImport(d *schema.ResourceData, m interface{}) ([]*schema.ResourceData, error) {
    apiClient := m.(*civogo.Client)

    // extract id and region from provided string
    parts := strings.Split(d.Id(), "_")
    if len(parts) != 2 {
        return nil, errors.New("To import a pool, use the format {pool_id}_{region}")
    }

    poolId := parts[0]
    region := parts[1]

    // Set new ID for resource
    d.SetId(poolId)

    // Set region
    d.Set("region", region)

    // Do the rest of the import

    ...
}