Java 13+ AWS Lambda Custom Runtime

About

This custom AWS Java runtime makes it easy to specify any JDK version (including pre-releases published by the AdoptOpenJDK project) and makes deploying the runtime to AWS as easy as running mvn install. The runtime uses the jlink utility of the Java Platform Module System to create a stripped-down (lean) build of the JDK. It also uses Application Class-Data Sharing, a feature introduced in Java 13. Both of these help to reduce the JDK startup time.

Supported Lambda Event Handler Methods

In order for the custom runtime to call your Lambda event handler it must use one of the following methods as it's entry point.

• public void handleRequest(InputStream input, OutputStream output)
• public void handleRequest(InputStream input, OutputStream output, String context)

The second method will be passed a JSON String containing an AWS Lambda Context object.

These are very general entry-points that can easily be used when handling a specific JSON object (such as an APIGatewayProxyRequestEvent when using a Lambda function connected to an API Gateway) like so (using the jackson-jr library in this example):

public class LambdaEventHandler {
    public void handleRequest(InputStream input, OutputStream output, String context) throws IOException {
        APIGatewayProxyRequestEvent event = JSON.std.with(JSON.Feature.FAIL_ON_UNKNOWN_BEAN_PROPERTY, false)
                .beanFrom(APIGatewayProxyRequestEvent.class, input);
        Context context = JSON.std.with(JSON.Feature.FAIL_ON_UNKNOWN_BEAN_PROPERTY, false)
                .beanFrom(LambdaContext.class, contextJson);
        // Do something with event.getHeaders(), event.getBody(), event.getQueryStringParameters(), etc.
        // Return a APIGatewayProxyResponseEvent:
        APIGatewayProxyResponseEvent response = handleEvent(event.getBody());
        JSON.std.write(response, output);
    }
}

It is not necessary for the Lambda function to extend or implement any abstract class or interface.

Change JDK Version

The version of the JDK that will be used when deploying the runtime can be configured using the following properties in the configuration of gmavenplus-plugin in pom.xml:

<configuration>
   <properties>
     <repoVersion>19</repoVersion>
     <type>jdk</type>
     <arch>x64</arch>
     <os>linux</os>
     <impl>hotspot</impl>
     <releaseDate>2022-02-03-05-48</releaseDate>
   </properties>
</configuration>

Configure AWS Region

You can change the region that will be used for deploying the runtime by the awsRegion property of the gmavenplus-plugin configuration in pom.xml:

<configuration>
   <properties>
      <awsRegion>US_WEST_2</awsRegion>
      <!-- etc. -->
   </properties>
    <!-- etc. -->
</configuration>

Configure jlink Modules

Because we use jlink to create a stripped-down build of the JDK it is necessary to explicitly add the modules that are needed to run whatever lambdas will be run by the custom runtime. The list of modules can be specified by the jlinkModules property of the gmavenplus-plugin configuration in pom.xml:

<configuration>
   <properties>
      <jlinkModules>java.net.http,java.desktop,java.logging,java.naming,java.sql,java.xml,org.slf4j,org.slf4j.simple</jlinkModules>
      <!-- etc. -->
   </properties>
    <!-- etc. -->
</configuration>

Deploy Runtime to AWS

The first step to deploy the runtime to your AWS account is to clone this repo. Next, the custom runtime is deployed to AWS by running mvn install. To build the runtime without deploying to AWS (for testing, for example) run mvn package.

mvn-assembly-plugin

Make sure you have copied src/assembly/lambda_deployment_package_assembly.xml to your project.

TODO (Public Consumption)

• Make it possible to supply a custom name for the published runtime.
• Allow for specifying a release version, not just a release date.

Credit

This project was forked from andthearchitect/aws-lambda-java-runtime which provided a great starting point!