Lambda-Go

This project is an attempt to understand how the most popular serverless computing solution, Lambda functions, works.

It was inspired by the most popular solution, AWS Lambda.

How it works

At this stage, the project is a web server with two endpoints. Server port should be specified in the APP_SERVER_PORT environment variable.

Also we need to specify logs level in the APP_LOG_LEVEL environment variable.

For proper operation, the server must have access to the Docker daemon, which is used to deploy our functions in containers.

Create function

This endpoint allows us to upload an archive with Lambda function code to the server.

curl --location 'localhost:9000/lambda/{func_name}/create' \
--form 'file=@"/func.tar.gz"'

To create a new Lambda function, you need to specify the function name {func_name} and a tar.gz archive.

The archive must contain three files:

• main.go
• go.mod
• go.sum

The code for the application should look like the following:

package main

import (
	"bytes"
	"context"
	"encoding/json"
	"fmt"

	"github.com/ihippik/lambda-go/lambda"
)

type Data struct {
	Name string `json:"name"`
}

func hello(_ context.Context, data []byte) ([]byte, error) {
	var req Data

	if err := json.NewDecoder(bytes.NewReader(data)).Decode(&req); err != nil {
		return nil, err
	}

	return []byte(fmt.Sprintf("Hello %s!", req.Name)), nil
}

func main() {
	lambda.Start(hello)
}

All handlers must have the following signature:

type Handler func(ctx context.Context, payload []byte) ([]byte, error)

To prepare the archive, you can use the following command:

tar -czvf func.tar.gz main.go go.mod go.sum

Invoke function

This endpoint allows us to run a previously uploaded function {func_name}

curl --location 'localhost:9000/lambda/{func_name}/invoke' \
--header 'Content-Type: application/json' \
--data '{"name": "Ivan"}'