如果 aws_api_gateway_integration 是在 Terraform 模块中创建的,我如何在 aws_api_gateway_deployment 资源上填充 depends_on?

How do I populate the depends_on on an aws_api_gateway_deployment resource if the aws_api_gateway_integration was created in a Terraform module?

aws_api_gateway_deployment 的 Terraform 文档中说:

Note: Depends on having aws_api_gateway_integration inside your rest api (which in turn depends on aws_api_gateway_method). To avoid race conditions you might need to add an explicit depends_on = ["aws_api_gateway_integration.name"].

我的 aws_api_gateway_deployment 资源位于根模块中,但大部分 aws_api_gateway_integration 是在子模块中创建的(这是我创建的本地模块)。

我的理解是您不能从模块中导出资源。

文件夹结构为:

 - main.tf <-- contains the aws_api_gateway_rest_api and aws_api_gateway_deployment and uses the service_func_lambda module multiple times
 - modules/
   - service_func_lambda/
     - main.tf <-- contains the aws_api_gateway_integration and other bits such as aws_api_gateway_method and aws_api_gateway_resource

如何从调用根模块引用在模块内部创建的 aws_api_gateway_integration

您不能依赖另一个模块中的资源。您可以通过引用该模块的输出来创建对整个模块的隐式依赖。

我认为您可以使用 null_resource(尽管可能有更好的方法)。像这样创建一个空资源,然后让你的 aws_api_gateway_deployment 依赖于它:

resource "null_resource" "depend_on_module" {
  triggers {
    service_func_lambda_module_output = "${module.service_func_for_lambda.some_output}"
  }
}

所以最后我让 aws_api_gateway_deployment 依赖于整个模块。这似乎工作得很好:

resource "aws_api_gateway_deployment" "api_gw_deploy" {

  depends_on = [
    "module.user_func_list",
    "module.user_func_create",
    "module.resource_func_list",
  ]

  rest_api_id = "${aws_api_gateway_rest_api.main_api_gw.id}"
  stage_name  = "live"
}