Building PHP Lambda functions with Docker container images

At re:Invent 2020, AWS announced that you can package and deploy AWS Lambda functions as container images. Packaging AWS Lambda functions as container images brings some notable benefits for developers running custom runtimes, such as PHP. This blog post explains those benefits and shows how to use the new container image support for Lambda functions to build serverless PHP applications.

Overview

Many PHP developers are familiar with building applications as containers to create a portable artifact for easier deployment. Packaging applications as containers helps to maintain consistent PHP versions, package versions, and configurations settings across multiple environments.

The new container image support for Lambda allows you to use familiar container tooling to build your applications. It also allows you to transition your applications into a serverless event-driven model. This brings the benefits of having no infrastructure to manage, automated scalability and a pay-per-use billing.

The advantages of an event-driven model for PHP applications are explained across the blog series “The serverless LAMP stack”. It explores the concepts, methods, and reasons for creating serverless applications with PHP. The architectural patterns and service limits in this blog series apply to functions packaged using both container image and zip archive formats, with some key exceptions:

Custom runtimes with container images

For custom runtimes such as PHP, Lambda provides base images containing the required Amazon Linux or Amazon Linux 2 operating system. Extend this to include your own runtime by implementing the Lambda Runtime API in a bootstrap file.

Before container image support for Lambda, a custom runtime is packaged using the .zip format. This required the developer to:

1. Set up an Amazon Linux environment compatible with the Lambda execution environment.
2. Install compilation dependencies and compile a version of PHP.
3. Save the compiled PHP binary together with a bootstrap file and package as a .zip.
4. Publish the .zip as a runtime layer.
5. Add the runtime layer to a Lambda function.

Any edits to the custom runtime such as new packages, PHP versions, modules, or dependences require the process to be repeated. This process can be time consuming and prone to error.

Creating a custom PHP runtime using the new container image support for Lambda can simplify changing the runtime environment. Dockerfiles allow you to have a fully scripted, faster, and portable build process without setting up an Amazon Linux environment.

This GitHub repository contains a custom PHP runtime for Lambda functions packaged as a container image. The following Dockerfile uses the base image for Amazon Linux provided by AWS. The instructions perform the following:

• Install system-wide Linux packages (zip, curl, tar).
• Download and compile PHP.
• Download and install composer dependency manager and dependencies.
• Move PHP binaries, bootstrap, and vendor dependencies into a directory that Lambda can read from.
• Set the container entrypoint.

#Lambda base image Amazon Linux
FROM public.ecr.aws/lambda/provided as builder 
# Set desired PHP Version
ARG php_version="7.3.6"
RUN yum clean all && \
    yum install -y autoconf \
                bison \
                bzip2-devel \
                gcc \
                gcc-c++ \
                git \
                gzip \
                libcurl-devel \
                libxml2-devel \
                make \
                openssl-devel \
                tar \
                unzip \
                zip

# Download the PHP source, compile, and install both PHP and Composer
RUN curl -sL https://github.com/php/php-src/archive/php-${php_version}.tar.gz | tar -xvz && \
    cd php-src-php-${php_version} && \
    ./buildconf --force && \
    ./configure --prefix=/opt/php-7-bin/ --with-openssl --with-curl --with-zlib --without-pear --enable-bcmath --with-bz2 --enable-mbstring --with-mysqli && \
    make -j 5 && \
    make install && \
    /opt/php-7-bin/bin/php -v && \
    curl -sS https://getcomposer.org/installer | /opt/php-7-bin/bin/php -- --install-dir=/opt/php-7-bin/bin/ --filename=composer

# Prepare runtime files
# RUN mkdir -p /lambda-php-runtime/bin && \
    # cp /opt/php-7-bin/bin/php /lambda-php-runtime/bin/php
COPY runtime/bootstrap /lambda-php-runtime/
RUN chmod 0755 /lambda-php-runtime/bootstrap

# Install Guzzle, prepare vendor files
RUN mkdir /lambda-php-vendor && \
    cd /lambda-php-vendor && \
    /opt/php-7-bin/bin/php /opt/php-7-bin/bin/composer require guzzlehttp/guzzle

###### Create runtime image ######
FROM public.ecr.aws/lambda/provided as runtime
# Layer 1: PHP Binaries
COPY --from=builder /opt/php-7-bin /var/lang
# Layer 2: Runtime Interface Client
COPY --from=builder /lambda-php-runtime /var/runtime
# Layer 3: Vendor
COPY --from=builder /lambda-php-vendor/vendor /opt/vendor

COPY src/ /var/task/

CMD [ "index" ]

To deploy this Lambda function, follow the instructions in the GitHub repository.

All runtime-related instructions are saved in the Dockerfile, which makes the custom runtime simpler to manage, update, and test. You can add additional Linux packages by appending to the yum install command. To install alternative PHP versions, change the php_version argument. Import additional PHP modules by adding to the compile command.

View the complete application in the following file tree:

project/
┣ runtime/
┃ ┗ bootstrap
┣ src/
┃ ┗ index.php
┗ Dockerfile

The Lambda function code is stored in the src directory in a file named index.php. This contains the Lambda function handler “index()”.

A bootstrap file is in the ‘runtime’ directory. This uses the Lambda runtime API to communicate with the Lambda execution environment.

The shebang hash sequence at the beginning of the bootstrap script instructs Lambda to run the file with the PHP executable, set by the Dockerfile.

All environment variables used in the bootstrap are set by the Lambda execution environment when running in the AWS Cloud. When running locally, the Lambda Runtime Interface Emulator (RIE) sets these values.

#!/var/lang/bin/php

Testing locally with the Lambda RIE

Using container image support for Lambda makes it easier for PHP developers to test Lambda functions locally. The previous container image example builds from the Lambda base image provided by AWS. This base image contains the Lambda RIE.

This is a proxy for Lambda’s Runtime and Extensions APIs. It acts as a lightweight web server that converts HTTP requests to JSON events and maintains functional parity with the Lambda Runtime API in the AWS Cloud. This allows developers to test functions locally using familiar tools such as cURL and the Docker CLI.

1. Build the previous custom runtime image using the Docker build command:
   docker build -t phpmyfuntion .
2. Run the function locally using the Docker run command, bound to port 9000:
   docker run -p 9000:8080 phpmyfuntion:latest
3. This command starts up a local endpoint at:
   localhost:9000/2015-03-31/functions/function/invocations
4. Post an event to this endpoint using a curl command. The Lambda function payload is provided by using the -d flag. This is a valid Json object required by the Runtime Interface Emulator:
   curl "http://localhost:9000/2015-03-31/functions/function/invocations" -d '{"queryStringParameters": {"name":"Ben"}}'
5. A 200 status response is returned:
   

Building web applications with Bref container images

Bref is an open source runtime Lambda layer for PHP. Using the bref-fpm layer, you can build applications with traditional PHP frameworks such as Symfony and Laravel. Bref’s implementation of the FastCGI protocol returns an HTTP response instead of a JSON response. When using the zip archive format to package Lambda functions, Bref’s custom runtime is provided to the function as a Lambda layer. Functions packaged as container images do not support adding Lambda layers to the function configuration. In addition to runtime layers, Bref also provides a number of Docker images. These images use the Lambda runtime API to form a runtime interface client that communicates with the Lambda execution environment.

The following example shows how to compose a Dockerfile that uses the bref php-74-fpm container image:

# Uses PHP 74-fpm.0, as the base image
FROM bref/php-74-fpm
# download composer for dependency management
RUN curl -s https://getcomposer.org/installer | php
# install bref using composer
RUN php composer.phar require bref/bref
# copy the project files into a Location that the Lambda service can read from
COPY . /var/task
#set the function handler entry point
CMD _HANDLER=index.php /opt/bootstrap

1. The first line sets the base image to use bref/php-74-fpm.
2. Composer, a dependency manager for PHP is installed.
3. Composer’s require command is used to add the bref package to the composer.json file.
4. The project files are then copied into the /var/task directory, where the function code runs from.
5. The function handler is set along with Bref’s bootstrap file.

The steps to build and deploy this image to the Amazon Elastic Container Registry are the same for any runtime, and explained in this announcement blog post.

Conclusion

The new container image support for Lambda functions allows developers to package Lambda functions of up to 10 GB in size. Using the container image format and a Dockerfile can make it easier to build and update functions with custom runtimes such as PHP.

Developers can include specific language versions, modules, and package dependencies. The Amazon Linux and Amazon Linux 2 base images give developers a starting point to customize the runtime. With the Lambda Runtime Interface Emulator, it’s simpler for developers to test Lambda functions locally. PHP developers can use existing third-party images, such as bref-fpm, to create web applications in a single Lambda function.

Visit serverlessland.com for more information on building serverless PHP applications.