Skip to content

Template for a Go service that has HTTP server, worker, and job components. Includes local setups for multiple databases via Docker Compose and deployment manifests for Kubernetes via Helm

License

zephinzer/template-go-service

Repository files navigation

Template Go Service

This is a template repository that provides boilerplate code for a service written in Go.

Supported data stores

  • Kafka
  • MongoDB
  • MySQL
  • NATS
  • PostgreSQL
  • Redis

Supported deployment methods

  • Kubenretes via Helm

Boilerplate information

Directory structure

Directory path Description
/.data Persistent data storage
/bin Contains built binaries
/cmd Contains command entrypoints
/deploy Contains deployment manifests
/internal Contains app-related packages
-> ./api Contains HTTP-related packages and defines mainly routing
-> ./constants Contains build configurations and app-wide constants
-> ./database Contains database connection management code
-> ./docs Contains Swagger documentation
-> ./example Contains an example endpoint that connects to the HTTP API
-> ./server Contains the server instance code
/tests Contains test artifacts
/vendor Contains dependenices

CLI information

To test the CLI you can run go run ./cmd/app, help information should be available.

The following commands are included in the boilerplate which you can remove if not applicable to your project:

  • app debug kafka - tests the connection to Kafka (run make kafka-jks and make start-kafka before this)
  • app debug mongo - tests the connection to MongoDB (run make start-mongo before this)
  • app debug mysql - tests the connection to MySQL (run make start-mysql before this)
  • app debug nats - tests the connection to NATS (run make start-nats before this)
  • app debug postgres - tests the connection to PostgreSQL (run make start-postgres before this)
  • app debug redis - tests the connection to Redis (run make start-redis before this)
  • app start job - sample command to start a one-time job
  • app start server - sample command to start the server
  • app start worker - sample command to start a worker service

Notable packages and software used

Go

  1. cobra for structuring the app's CLI invocations and flags configuration
  2. viper for managing configuration on a service level
  3. gofiber for the HTTP server and HTTP API scaffolding
  4. swaggo to generate Swagger documentation

Tooling

  1. Docker for build/release packaging
  2. Docker Compose for bringing up development services
  3. KinD for bringing up a local Kubernetes cluster
  4. nk for generating NATS nkeys
  5. Make for development operations recipes

Supported data stores

Queues/Message brokers

  1. Kafka
  2. NATS for queueing/message brokering

Relational databases

  1. MySQL for RDBMS usage
  2. PostgreSQL for RDBMS usage

NoSQL/document databases/Key stores

  1. MongoDB for NoSQL usage
  2. Redis for caching

SDLC Usage

Development operations

This repository uses Makefile for storing development operations. Included operations are:

  • make binary builds the binary and stores it in ./bin/
  • make deploy-kind deploys the application onto a local Kubernetes cluster
  • make deploy-k8s deploys the application onto the currently selected Kubernetes cluster
  • make deps pulls in the dependencies into ./vendor/
  • make docs generates all documentation
  • make docs-swaggo generates just the Swagger documentation using swaggo
  • make image builds the Docker image
  • make install-swaggo installs swaggo at the latest version
  • make kafka-jks generates the keys and certificates required for Kafka security
  • make kind-load loads the image into the local Kubernetes cluster
  • make nats-nkey creates a new nkey for use with NATS
  • make publish-image publishes the image in it's current form using a concatenation of ${GIT_BRANCH} and ${GIT_COMMIT_SHA} separated by a - as the image tag
  • make start starts the HTTP server using defaults
  • make start-kafka starts a local Kafka instance via Docker Compose
  • make start-kind starts a local Kubernetes cluster with KinD
  • make start-mongo starts a local MongoDB instance
  • make start-mysql starts a local MySQL instance
  • make start-nats starts a local NATS instance
  • make start-postgres starts a local PostgreSQL instance
  • make start-redis starts a local Redis instance
  • make test runs tests on the application and produces coverage artifacts at ./tests/

To configure the Makefile recipes, create a new Makefile.properties in the root of this repository. Configurable values can be found in the Makefile above the line -include Makefile.properties.

Some common configurations are documented below. Insert the code snippets into Makefile.properties.

See the ./Makefile for full details

Database connections

Sample connection code with test pings can be found in ./cmd/app/commands/debug.

All database packages at ./internal/database expose the global interface:

  • Db(name ...string): Use this to retrieve a database connection. If name is specified, the named connection will be returned.
  • Init(opts database.ConnectionOpts, name ...string): Use this to initialise a database connection. If name is specified, a named connection will be created and added to a map of connections which you can retrieve using Db(connectionName).
  • Close(name ...string): Use this to close connections you no longer need. If name is specified, a named connection will be closed and removed from the connection map. You need to call this before you Init() another connection with the same name.

Named connections can be used to implement logical tenant separations at the connection level. For example for B2B businesses where each tenant has a separate database instance, a new connection for tenantA can be initialised using Init(opts, "tenantA") and retrieved using Db("tenantA").

Building on top of the boilerplate

Changing the application name

Insert the following into Makefile.properties:

APP_NAME := "app2"

You will also need to:

  1. Rename the module in go.mod to app2 and run go mod vendor
  2. Rename the chart in ./deploy/charts/app/Chart.yaml to app2
  3. Rename the directory ./deploy/charts/app to ./deploy/charts/app2
  4. Rename the directory in ./cmd/app to ./cmd/app2

Changing the Docker image URL

Insert the following into Makefile.properties:

IMAGE_REGISTRY := "docker.registry.domain.com"
IMAGE_PATH := "your-org/your-app"
IMAGE_TAG := "main-123456"

This example will cause images to be tagged and pushed to docker.registry.domain.com/your-org/your-app:main-123456

Adding a new functional domain

Domains should be in their own internal package in ./internal/. A package should contain both the controllers and the HTTP interfaces exposed with a reference performed in ./internal/api/

An example with both controllers and HTTP interfaces can be found in ./internal/example for an example domain.

Pre-production checklist

This repository being a boilerplate contains some redunancies like database code which might not be used depending on your service. Before going to production:

  • Helm chart is name appropriately
  • Service is named appropriately
  • Remove the ./cmd/app/commands/debug/* directories that aren't used. Eg if MongoDB is not used, remove ./cmd/app/commands/debug/mongo and remove the reference to it in ./cmd/app/commands/debug/debug.go. This way, code for handling that database type will not be included in your binary.
  • Disable Swagger in production (unless you want it in production)
  • Generate fresh keys on a privileged system and delete any keys that come with the boilerplate

Licensing

See ./LICENSE. Basically, do whatever you want with this.

About

Template for a Go service that has HTTP server, worker, and job components. Includes local setups for multiple databases via Docker Compose and deployment manifests for Kubernetes via Helm

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published