AWS Lambda for SparkleFormation

Lets make lambda functions easier to manage in CloudFormation!

This SparkleFormation Callback adds a new helper method to AWS based SparkleFormation templates called lambda!. This helper method will insert an AWS::Lambda::Function resource into the template using source files contained within configured directories.

• Individual source files for lambda functions
• Automatic resource creation within templates
• Automatic code setup for resource
• Acceptable functions will be defined inline
• S3 storage will be used when inline is unacceptable
• S3 versioning will be used when bucket configured for versioning
• Automatic asset builds (for java8 runtime targets)

First add the sfn-lambda gem to the local bundle (in the ./Gemfile):

group :sfn do
  gem 'sfn-lambda'
end

Now enable the sfn-lambda callback in the .sfn configuration file:

Configuration.new do
  ...
  callbacks do
    require ['sfn-lambda']
    default ['lambda']
  end
  ...
end

NOTE: If using the java8 runtime for lambda functions, maven must be installed with mvn being available within the user's PATH.

By default the sfn-lambda callback will search the ./lambda directory for lambda function files. A custom directory path can be used by modifying the configuration:

Configuration.new do
  lambda do
    directory './my-lambdas'
  end
end

By default the sfn-lambda callback will use the bucket name provided by the nesting_bucket configuration item. This can be customized to use a different bucket by modifying the configuration:

Configuration.new do
  lambda do
    upload do
      bucket 'my-custom-bucket'
    end
  end
end

The path of lambda function files is important. The path is used to determine the proper handler for running the lambda function, as well as providing the identifier to reference the function. The path structure is as follows:

./lambda/RUNTIME/FUNCTION_NAME.extension

The RUNTIME defines the runtime used for handling the lambda function. At the time of writing this, that value can be one of:

• nodejs
• nodejs4.3
• java8
• python2.7

NOTE: Runtime values are not validated which allows new runtimes to be used as they are made available.

The FUNCTION_NAME is used for two purposes:

1. It identifies the function name lambda should use
2. It is used in combination with the RUNTIME to identify the lambda in the template

Using the python example described within the lambda documentation:

• http://docs.aws.amazon.com/lambda/latest/dg/python-programming-model-handler-types.html

we can define our handler code:

• ./lambda/python2.7/my_function.py

def my_handler(event, context):
    message = 'Hello {} {}!'.format(event['first_name'], event['last_name'])
    return {
        'message' : message
    }

Now, using a new helper method, lambda resources can be created within a SparkleFormation template using the newly created file:

• ./sparkleformation/lambda_test.rb

SparkleFormation.new(:lambda_test) do
  lambda!(:my_function, :handler => :my_handler)
end

If the handler argument is not specified the default value is 'handler'.

SparkleFormation.new(:lambda_test) do
  lambda!(:my_function)
end

When the template is printed a lambda resource is shown with the function properly inlined:

$ sfn print --file lambda_test
{
  "Resources": {
    "MyHandlerLambdaFunction": {
      "Type": "AWS::Lambda::Function",
      "Properties": {
        "Handler": "index.my_handler",
        "Runtime": "python2.7"
        "FunctionName": "my_function",
        "ZipFile": "def my_handler(event, context):\n    message = 'Hello {} {}!'.format(event['first_name'], event['last_name'])\n    return {\n        'message' : message\n    }\n\n"
      }
    }
  }
}

If the name of a lambda function is shared across multiple runtimes, the desired runtime can be specified within the call:

SparkleFormation.new(:lambda_test) do
  lambda!(:my_function, :handler => :my_handler, :runtime => 'python2.7')
end

If a lambda function is to be used for creating multiple resources within a template, a custom name can be added as well:

SparkleFormation.new(:lambda_test) do
  lambda!(:my_function, :first, :handler => :my_handler, :runtime => 'python2.7')
  lambda!(:my_function, :second, :handler => :my_handler, :runtime => 'python2.7')
end

When the size of the lambda function is greater than the defined max size (4096 default), the function will be stored on S3. If the bucket configured for storage has versioning enabled, versioning information will be automatically set within the resource. If no versioning information is available, a checksum will be attached to the generated key name.

For lambda functions utilizing the java8 runtime, the sfn-lambda callback will behave slightly different. When discovering available lambda functions, the directory names under the ./lambda/java8 directory will be used. This allows for the collection of required files to be stored within the directory.

Using the example here: http://docs.aws.amazon.com/lambda/latest/dg/java-create-jar-pkg-maven-no-ide.html

The defined directory structure would be:

$ cd ./lambda
$ tree
.
|____java8
| |____hello
| | |____src
| | | |____main
| | | | |____java
| | | | | |____example
| | | | | | |____Hello.java
| | |____pom.xml

When the hello lambda function is used within a template, sfn-lambda will automatically generate the required jar file using Maven and store the resulting asset on S3.

NOTE: Maven is required to be installed when using the java8 runtime

• Repository: https://github.com/sparkleformation/sfn-lambda
• IRC: Freenode @ #sparkleformation