@sls-next/lambda-at-edge
Library to build and deploy Next.js apps for AWS Lambda@Edge
This library uses the handlers provided by @sls-next/core
and wraps them with a Lambda@Edge/CloudFront-compatible layer.
Usage
const path = require('path');
const { Builder } = require("@sls-next/lambda-at-edge");
const nextConfigPath = '/path/to/my/nextapp';
const outputDir = path.join(nextConfigPath, ".serverless_nextjs");
const builder = new Builder(
nextConfigPath,
outputDir,
{
cmd: './node_modules/.bin/next',
cwd: process.cwd(),
env: {},
args: ['build'],
minifyHandlers: true,
// it is recommended to let your CF distribution do the compression as per the docs - https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/ServingCompressedFiles.html
// however there have been issues in the past where CF doesn't compress lambda@edge responses, so we provide our own implementation in case is needed
enableHTTPCompression: false
}
);
await builder.build()
.then(() => {
console.log("Application built successfully!");
})
.catch((e) => {
console.log("Could not build app due the exception: ", e);
process.exit(1);
});
You can configure more options regarding building process. Configurable inputs you can find in 'build.ts' file ('packages/libs/lambda-at-edge/src/build.ts'). If you want to see debug logs during building, use 'await builder.build(true)' instead. After running the above, the output directory will contain the Lambda@Edge handlers necessary to server side render at the edge.
/dir/to/my/next-app/.serverless_nextjs/
> default-lambda
> manifest.json
> routes-manifest.json
> prerender-manifest.json
> pages/
> index.js # handler
> api-lambda
> manifest.json
> routes-manifest.json
> pages/api/
> index.js # handler
> image-lambda
> manifest.json
> routes-manifest.json
> images-manifest.json
> node_modules/...
> index.js # handler
The handlers need to be attached to the origin-request
trigger of CloudFront.
The api-lambda
edge function should be attached to a CloudFront behaviour that only triggers in the event of /api/*
requests.
The image-lambda
edge function should be attached to a CloudFront behaviour that only triggers in the event of _next/image*
requests.
For full usage docs, please refer to (TBA).
Architecture
Once built and packaged, the app consists of the following components:
-
default-lambda
(v2): handles page and API requests. -
default-lambda
(v1): handles page requests only. -
api-lambda
(legacy v1 handlers only): handles API requests. -
regeneration-lambda
: handles regeneration requests used for ISR. -
image-lambda
: handles image optimization requests. -
assets
: all static assets used by your app.
Infrastructure
You will need the following infrastructure to deploy your app:
- AWS Lambda@Edge
- AWS CloudFront
- AWS API Gateway
- AWS S3 Bucket
- AWS SQS Queue (if you are using ISR)
- additional roles, permissions, etc.
Deployment
Currently, you will need to deploy via the Serverless Components deployer, @sls-next/serverless-component
. We also provide a CDK construct at @sls-next/nextjs-cdk-construct
.
If you'd like to write your own custom deployment logic, please see the CDK construct or legacy Serverless Components deployer to see an example of all the infrastructure you need to setup.
Limitations
- Lambda@Edge limitations apply: https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/edge-functions-restrictions.html. Most notably, there is a 1 MB response size limit, which could especially affect static files.
- CloudFront limitations apply: https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/cloudfront-limits.html
- The image handler only serves image optimization requests. It cannot redirect, rewrite or add headers (yet).
- In v1 handlers (legacy), the default, API and image handlers are separate, so you cannot rewrite from default routes -> image routes nor rewrite between default routes and API routes.
- In v2 handlers, the default and image handlers are separate, so you cannot rewrite from default routes -> image routes.
Acknowledgements
Special thanks for Daniel Conde Marin for the initial implementation.