⚡️Serverless Framework Go Plugin

serverless-plugin-go is a Serverless Framework plugin that compiles Go functions on the fly. You don't need to do it manually before serverless deploy. Once the plugin is installed it will happen automatically.

Features

• Concurrent compilation happens across all CPU cores.
• Support for both serverless deploy and serverless deploy function commands.
• Support for serverless invoke local command.
• Additional command serverless go build.

Install

1. Install the plugin

   npm i --save-dev serverless-plugin-go

2. Add it to your serverless.yaml

   plugins:
     - 'serverless-plugin-go'

3. Replace every Go function's handler with *.go file path or a package path. E.g.

   functions:
     example:
       runtime: 'provided.al2'
       handler: 'functions/example/main.go' # or just functions/example

Configuration

Default values:

custom:
  go:
    # folder where go.mod file lives, if set `handler` property should be set relatively to that folder
    baseDir: '.'

    # Final target destination of the binary outputs
    binDir: '.bin'

    # Compile command that should be used to build binaries
    cmd: 'go build -ldflags="-s -w"'

    # If enabled, builds function every directory (useful for monorepo where go.mod is managed by each function
    monorepo: false

    # Number of concurrent process used to compile functions, you can fine tune this number to your needs.
    # Alternatively you can use the env var `SP_GO_CONCURRENCY` to override this configuration
    concurrency: 5

    # Environment variables used to compile binaries
    env:
      GOOS: 'linux' # Default compile OS
      CGO_ENABLED: '0' # By default CGO is disabled

How does it work?

The plugin compiles every Go function defined in serverless.yaml into .bin directory. After that it internally changes handler so that the Serverless Framework will deploy the compiled file not the source file.

For every matched function it also overrides package parameter to the next config:

individually: true
artifact: <path to the generated zip with go binary>

How to run Golang Lambda on ARM?

Change architecture field from provider or function configuration to arm64:

provider:
  architecture: 'arm64'
  runtime: 'provided.al2'

# If you define architecture at function level, this will have preference instead the provider configuration

functions:
  test:
    architecture: 'arm64'
    handler: test/main.go

Caveats

This implementation doesn't allow to add any other files to the lambda artifact apart from the binary. To address this problem you should use Lambda Layers to allow your lambdas to hold external files.

Credits

This repo is forked from serverless-go-plugin created by mthenw and modified to fix some issues and reach the new AWS standards (go1.x provider deprecation).