Terraform中CloudFormation的Custom::LoadLambda等效资源咨询

Great question! Let's break this down clearly first: Custom::LoadLambda isn't a native CloudFormation resource—it's a custom CloudFormation resource powered by a backend Lambda function. Its exact job varies, but it’s typically used to handle custom workflows for loading Lambda code (like pulling from S3, dynamically packaging dependencies, or other specialized logic).

Terraform doesn’t have a direct 1:1 built-in equivalent for this custom resource, but you can replicate its functionality using Terraform’s native tools and resources. Here are the most common approaches based on what your Custom::LoadLambda does:

1. If it only loads pre-packaged code (local or S3) #

If your Custom::LoadLambda’s sole purpose is to point to a ready-to-use Lambda code zip (stored locally or in S3), you don’t need any custom tools—Terraform’s native aws_lambda_function resource handles this directly:

resource "aws_lambda_function" "my_lambda" {
  function_name = "my-target-function"
  role          = aws_iam_role.lambda_exec_role.arn
  runtime       = "python3.12" # Match your function's runtime
  handler       = "index.handler" # Match your function's handler

  # For a local code zip file
  filename = "./path/to/your/lambda-code.zip"

  # OR for an S3-hosted zip (uncomment below)
  # s3_bucket = "your-lambda-code-bucket"
  # s3_key    = "path/to/lambda-code.zip"
}

2. If it handles custom code packaging/dependency logic #

If your Custom::LoadLambda runs custom steps (like installing Python/Node.js dependencies, compiling code, or generating dynamic content), use Terraform’s local-exec provisioner to run scripts that replicate that logic, then pass the output to aws_lambda_function:

# First, run your custom packaging workflow
resource "null_resource" "package_lambda_code" {
  provisioner "local-exec" {
    command = <<EOT
      # Example: Install Python dependencies and package code
      mkdir -p ./lambda-build
      cp ./src/index.py ./lambda-build/
      pip install -r ./src/requirements.txt -t ./lambda-build/
      cd ./lambda-build && zip -r ../lambda-code.zip .
    EOT
  }
}

# Then create the Lambda function using the packaged zip
resource "aws_lambda_function" "my_lambda" {
  function_name = "my-target-function"
  role          = aws_iam_role.lambda_exec_role.arn
  runtime       = "python3.12"
  handler       = "index.handler"
  filename      = "./lambda-code.zip"

  # Ensure packaging finishes before creating the Lambda
  depends_on = [null_resource.package_lambda_code]
}

3. If you want to reuse the original custom logic #

If you need to keep using the exact backend Lambda function that powered your Custom::LoadLambda, you can invoke it directly from Terraform using the aws_lambda_invocation data source, then use its output to configure your Lambda function:

# Invoke the original Custom::LoadLambda backend function
data "aws_lambda_invocation" "load_lambda_config" {
  function_name = "your-custom-loadlambda-backend-function"

  # Pass the same parameters you used in CloudFormation
  input = jsonencode({
    S3Bucket = "your-code-bucket"
    S3Key    = "path/to/code.zip"
    # Add any other parameters your custom resource expects
  })
}

# Create the Lambda function using the result from the invocation
resource "aws_lambda_function" "my_lambda" {
  function_name = "my-target-function"
  role          = aws_iam_role.lambda_exec_role.arn
  runtime       = "python3.12"
  handler       = "index.handler"

  # Extract the S3 location from the Lambda invocation result
  s3_bucket = jsondecode(data.aws_lambda_invocation.load_lambda_config.result).S3Bucket
  s3_key    = jsondecode(data.aws_lambda_invocation.load_lambda_config.result).S3Key
}

Key Takeaway #

The exact Terraform replacement depends on what your Custom::LoadLambda actually does. Start by documenting the specific logic of that custom CloudFormation resource, then map it to Terraform’s native resources, provisioners, or data sources as shown above.

内容的提问来源于stack exchange，提问作者Simply Seth