Skip to content

Latest commit

 

History

History
executable file
·
241 lines (184 loc) · 8.65 KB

README.md

File metadata and controls

executable file
·
241 lines (184 loc) · 8.65 KB
Raw

Develop Body Mass Index(BMI) via go-chassis

Prerequisite

Install Golang development environment

  • Install Git, details can refer to Git Installing Guide

  • Install Golang 1.11+, details can refer to Golang Installing Guide

  • Install go-chassis(SDK) by executing the following commands.

    mkdir -p $GOPATH/src/github.com/go-chassis
cd $GOPATH/src/github.com/go-chassis
git clone https://github.com/go-chassis/go-chassis.git
export GO111MODULE=on
go mod vendor

Run Service Center

Service Center enables capabilities of service registration and service discovery in ServiceComb. It can run inside docker.

docker pull servicecomb/service-center
docker run -d -p 30100:30100 servicecomb/service-center:latest

Develop BMI microservice application in minutes

The workflow of BMI application have been introduced in Quick start. It contains two separate microservices:

  • BMI calculator service：provides computing services.

  • Web service：provides both user interface and gateway service.

Now we will introduce the detailed implementation of these two microservices. The full code is on github.

Implementation of calculator

The calculator service provides capability of calculating BMI. It contains three parts:

  • Business logic

  • Service registration and configuration

  • Start the framework

Business logic

Define the service structure

type CalculateBmi struct {
}

Implement detailed calculation according to the formula \(BMI=\frac{weight}{height^2}\).

func (c *CalculateBmi) BMIIndex(height, weight float64) (float64, error) {
	if height <= 0 || weight <= 0 {
		return 0, fmt.Errorf("Arugments must be above 0")
	}
	heightInMeter := height / 100
	bmi := weight / (heightInMeter * heightInMeter)
	return bmi, nil
}

Write handler function, which must be import the parameter type of restful.Context. restful.Context contains commonly used operations on Request and Response to http calls. Functions starts with Read is about read operations from Request , including ReadEntity , ReadHeader , ReadPathParameter , ReadBodyParameter , ReadRequest etc .; functions start with Writer is about write operation to Response , Including WriteHeader , WriteJson , WriteError and so on.

import (
  rf "github.com/go-chassis/go-chassis/server/restful"
)
	……
func (c *CalculateBmi) Calculate(b *rf.Context) {
	……					//Define the Response structure, omit here 
  heightStr := b.ReadQueryParameter("height")
  weightStr := b.ReadQueryParameter("weight")
  var height, weight, bmi float64
  var err error
	……  				// Convert form string to float, omit here
  if bmi, err = c.BMIIndex(height, weight); err != nil {
    errorResponse.Error = err.Error()
    b.WriteHeaderAndJson(http.StatusBadRequest, errorResponse, "application/json")
    return
  }
  result.Result = bmi
  result.CallTime = time.Now().String()
  b.WriteJson(result, "application/json")   
}

Specify the corresponding URL route

func (c *CalculateBmi) URLPatterns() []rf.Route {
  return []rf.Route{
    {http.MethodGet, "/calculator/bmi", "Calculate"},
  }
}

Service registration and configuration

After the completion of the preparation of the business logic code, the business logic needs to be registered to the Go-chassis framework, and at the time of registration, the service name, ID and other attributes can be specified simultaneously.

  chassis.RegisterSchema("rest",&CalculateBmi{})

In addition to some of the attributes specified in the code, more attributes are configured through the configuration file. The configuration file includes chassis.yaml and microservice.yaml , placed in the conf directory under the code directory. Among them, chassis.yaml contains public service properties, such as public AppId information, registry type information, registry address, service protocol, transmission protocol information; microservice.yaml is about the private properties of microservices, including the service name, version and so on.

chassis.yaml

APPLICATION_ID: bmi						   
cse:
  service:
    registry:
      address: http://127.0.0.1:30100    	
  protocols:
    rest:
      listenAddress: 0.0.0.0:8080

microservice.yaml

service_description:
  name: calculator						  
  version: 0.0.1

Start the framework

import (
  	"github.com/go-chassis/go-chassis"
	"github.com/go-chassis/go-chassis/core/lager"
)
	……
  if err := chassis.Init(); err != nil {           //Init the chassis framwork
    lager.Logger.Errorf("Init FAILED %s", err.Error())
    return
  }
  chassis.Run()                                     //Run the microservice

Implementation of web-app service

The web service provides both the user interface . It contains three parts:

  • Front-end static page

  • Request forward

  • Service configuration and startup

Front-end static page

Front-end static page is powered by Bootstrap , and through the golang official library http.ServeFile front-end static page will be displayed.

func BmiPageHandler(w http.ResponseWriter, r *http.Request) {
  	http.ServeFile(w,r,"external/index.html")
}

Request forward

The web-app service forwards requests to the calculator service via core.NewRestInvoker() when it receives a request from the front-end page. In the process of forwarding the call, the user does not need to know the specific address and port of the calculator service. The service discovery process is automatically completed by the go-chassis framework.

func BmiRequestHandler(w http.ResponseWriter, r *http.Request) {
  queries := r.URL.Query()
  heightStr := queries.Get("height")
  weightStr := queries.Get("weight")

  requestURI := fmt.Sprintf("cse://calculator/bmi?height=%s&weight=%s", heightStr, weightStr)
  restInvoker := core.NewRestInvoker()
  req, _ := rest.NewRequest("GET", requestURI)
  resp, _ := restInvoker.ContextDo(context.TODO(), req)

  w.Header().Set("content-type", "application/json")
  w.WriteHeader(resp.GetStatusCode())
  w.Write(resp.ReadBody())
}

Service configuration and startup

The configuration file for the web-app service also includes two yaml files, chassis.yaml and microservice.yaml , as follows: chassis.yaml

APPLICATION_ID: bmi						
cse:
  service:
    registry:
      address: http://127.0.0.1:30100    # ServiceCenter address

microservice.yaml

service_description:
  name: web-app							  
  version: 0.0.1

Unlike the calculator service , the web-app is a consumer-type service within the Go-chassis framework, so just invoke chassis.Init() to initialize the Go-chassis framework.

func main() {
  http.HandleFunc("/", BmiPageHandler)
  http.HandleFunc("/calculator/bmi", BmiRequestHandler)

  if err := chassis.Init(); err != nil {
    lager.Logger.Errorf("Init FAILED %s", err.Error())
    return
  }

  port := flag.String("port", "8889", "Port web-app will listen")
  address := flag.String("address", "0.0.0.0", "Address web-app will listen")
  fullAddress := fmt.Sprintf("%s:%s", *address, *port)
  http.ListenAndServe(fullAddress, nil)
}

Play with kubernetes

The docker builds & kubernetes orchestrations are already there in the Makefile. There are 3 stages from binary builds to kubernetes deployments:

  • build binaries: make $target, for example make calculator, this will build the calculator binary
  • build docker images: make docker.$target, for example make docker.calculator, this will build the calculator image
  • deploy image to k8s cluster: make k8s.$target, for example make k8s.calculator, this will deploy the calculator to the kubernetes cluster by a service and a deployment

Handle all by make:

  • make bin: build all the binaries
  • make docker: execute make bin, then build the docker images
  • make k8s: execute make docker, then deploy the newly built image to the kubernetes cluster

The k8s related make commands will call ./scripts/distribute-image.sh to dispatch docker images to k8s clusters. The script will try to find the local IP and the k8s cluster's nodes IP, and execute some SSH command on the nodes. So there are 2 points you'll take care of:

  1. Make sure ip command is executable, and modify the network adapter's name in ./scripts/distribute-image.sh:25, the default is wlp3s0, replace it with your local network card's name
  2. Make sure the current user has the permission to execute remote commands on the k8s cluster nodes.