Dapr 负载均衡

实现目标

Dapr 使用 mDNS 协议提供负载均衡，将Service02的请求平均分布到两个Service01的实例上。

创建API接口与服务

创建Service01的服务接口。
我们这里就只提供获取唯一的Guid接口服务。

[ApiController]
[Route("[controller]")]
public class WeatherForecastController : ControllerBase
{
    private static readonly string InstanceId = Guid.NewGuid().ToString();
    private readonly ILogger<WeatherForecastController> _logger;

    public WeatherForecastController(ILogger<WeatherForecastController> logger)
    {
        _logger = logger;
    }

    [HttpGet(Name = "guid")]
    public async Task<IActionResult> Get()
    {
        return Ok(new { Service = "Service01", InstanceId });
    }
}

创建Service02的服务接口。
首先安装Dapr.AspNetCore包，然后添加好Dapr依赖服务，并创建一个Guid的控制器。

builder.Services.AddControllers()
    .AddDapr();

[ApiController]
[Route("[controller]")]
public class GuidController : ControllerBase
{
    private readonly DaprClient _daprClient;

    public GuidController(DaprClient daprClient)
    {
        _daprClient = daprClient;
    }

    [HttpGet]
    public async Task<IActionResult> Hello()
    {
        // 通过dapr访问Service01的API
        var result = await _daprClient.InvokeMethodAsync<object>(HttpMethod.Get, "service01", "WeatherForecast");
        return Ok(result);
    }
}

运行Service01服务

首先启动我们的两个需要负载均衡的接口服务，运行的端口为6002与6001，在使用daprd命令注册到dapr中。

dotnet run service01 --urls=http://localhost:6002
dotnet run service01 --urls=http://localhost:6001

daprd --app-id service01 --app-port 6001 --dapr-http-port 5001  --dapr-grpc-port 4001 --metrics-port 9001
daprd --app-id service01 --app-port 6002 --dapr-http-port 5002  --dapr-grpc-port 4002 --metrics-port 9002

运行Service02服务

打开Service02项目的目录下，通过powershell执行下面的命令进行启动。

dapr run --dapr-http-port 5003 --app-port 6003 --app-id service02 -- dotnet run service02 --urls=http://localhost:6003

测试运行

http://localhost:5001/v1.0/invoke/service01/method/WeatherForecast
http://localhost:5002/v1.0/invoke/service01/method/WeatherForecast
# 测试是否达到了负载均衡
http://localhost:5003/v1.0/invoke/service02/method/guid

部署到k8s中

镜像打包与上传

docker build -t aidasi/dapr_service01:v1 -f ./DaprLoadBalancer/Dockerfile .
docker build -t aidasi/dapr_service02:v1 -f ./DaprLoadBalancerGateway/Dockerfile .
# push
docker push aidasi/dapr_service01:v1
docker push aidasi/dapr_service02:v1

kubernetes 部署相关服务

在确保有dapr的环境下进行部署dapr_service01和dapr_service02服务，并将需要负载均衡的dapr_service01副本数设置为2.

vim service01.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: service01
  labels:
    app: service01
spec:
  replicas: 2 # 2个副本
  selector:
    matchLabels:
      app: service01
  template:
    metadata:
      labels:
        app: service01
      annotations:
        dapr.io/enabled: "true" # 开启Dapr边车
        dapr.io/app-id: "service01"
        dapr.io/app-port: "80"
    spec:
      containers:
      - name: service01
        image: docker.io/aidasi/dapr_service01:v1

kubectl apply -f service01.yaml

创建Service02的yaml文件

vim service02.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: service02
  labels:
    app: service02
spec:
  replicas: 1
  selector:
    matchLabels:
      app: service02
  template:
    metadata:
      labels:
        app: service02
      annotations:
        dapr.io/enabled: "true"  # 开启Dapr边车
        dapr.io/app-id: "service02"
        dapr.io/app-port: "80"
    spec:
      containers:
      - name: service02
        image: docker.io/aidasi/dapr_service02:v1
---
apiVersion: v1
kind: Service
metadata:
  name: service02
spec:
  type: ClusterIP
  selector:
    app: service02
  ports:
    - port: 80
      targetPort: 80

kubectl apply -f service02.yaml

测试

# 查看资源创建情况
kubectl get pod,svc
# 访问guid
curl http://10.110.174.130/guid
curl http://10.110.174.130/guid