1. 程式人生 > 其它 >Go gRPC-超時設定

Go gRPC-超時設定

Go gRPC-超時設定

一、前言

gRPC預設的請求的超時時間是很長的,當你沒有設定請求超時時間時,所有在執行的請求都佔用大量資源且可能執行很長的時間,導致服務資源損耗過高,使得後來的請求響應過慢,甚至會引起整個程序崩潰。

為了避免這種情況,我們的服務應該設定超時時間。前面的提到,當客戶端發起請求時候,需要傳入上下文context.Context,用於結束超時取消的請求。

如何設定gRPC請求的超時時間。

原始碼

二、新建proto檔案

syntax = "proto3";// 協議為proto3

//path 表示生成的go檔案的存放地址,會自動生成目錄的。
//name 表示生成的go檔案所屬的包名
//option go_package = "path;name";


option go_package = "./;proto";
package proto;

//  生成pb.go命令:  protoc -I ./ --go_out=plugins=grpc:.\06deadlines\proto\  .\06deadlines\proto\simple.proto
// 定義我們的服務(可定義多個服務,每個服務可定義多個介面)
service Simple{
  rpc Route (SimpleRequest) returns (SimpleResponse){};
}

// 定義傳送請求資訊
message SimpleRequest{
  // 定義傳送的引數,採用駝峰命名方式,小寫加下劃線,如:student_name
  // 引數型別 引數名 標識號(不可重複)
  string data = 1;
}

// 定義響應資訊
message SimpleResponse{
  // 定義接收的引數
  // 引數型別 引數名 標識號(不可重複)
  int32 code = 1;
  string value = 2;
}


// 指令編譯方法,進入專案所在目錄,執行
go-grpc-example> protoc -I ./ --go_out=plugins=grpc:.\06deadlines\proto\  .\06deadlines\proto\simple.proto

二、客戶端請求設定超時時間

1.把超時時間設定為當前時間+3秒

clientDeadline := time.Now().Add(time.Duration(3 * time.Second))
ctx, cancel := context.WithDeadline(ctx, clientDeadline)
defer cancel()

2.響應錯誤檢測中新增超時檢測

// 傳入超時時間為3秒的ctx
res, err := grpcClient.Route(ctx, &req)
if err != nil {
    //獲取錯誤狀態
    statu, ok := status.FromError(err)
    if ok {
        //判斷是否為呼叫超時
        if statu.Code() == codes.DeadlineExceeded {
            log.Fatalln("Route timeout!")
        }
    }
    log.Fatalf("Call Route err: %v", err)
}
// 列印返回值
log.Println(res.Value)

完整的client.go程式碼

package main

import (
	"context"
	pb "go-grpc-example/06deadlines/proto"
	"log"
	"time"

	"google.golang.org/grpc"
	"google.golang.org/grpc/codes"
	"google.golang.org/grpc/credentials/insecure"
	"google.golang.org/grpc/status"
)

/*
@author RandySun
@create 2022-05-08-15:53
*/
const (
	// Address 監聽地址
	Address string = ":8001"
)

var grpcClient pb.SimpleClient

//
//  route
//  @Description:設定超時時間
//  @param ctx
//  @param deadlines
//
func route(ctx context.Context, deadlines time.Duration) {
	// 設定超時時間
	clientDeadline := time.Now().Add(time.Duration(deadlines * time.Second))
	ctx, cancel := context.WithDeadline(ctx, clientDeadline)
	defer cancel()
	// 建立傳送結構體
	req := pb.SimpleRequest{
		Data: "grpc",
	}
	// 呼叫 Route 方法 同時傳入context.Context,  在有需要時可以讓我們改變RPC的行為,比如超時/取消一個正在執行的RPC
	res, err := grpcClient.Route(ctx, &req)
	// 列印返回直
	log.Println("服務的返回響應data:", res)
	if err != nil {

		// 獲取錯誤狀態
		statu, ok := status.FromError(err)

		if ok {
			// 判斷是否為呼叫超時
			if statu.Code() == codes.DeadlineExceeded {
				log.Fatalln("Route timeout!")
			}
		}
		log.Fatalln("Call Route err:", err)

	}

}

func main() {
	// 連線伺服器
	conn, err := grpc.Dial(Address, grpc.WithTransportCredentials(insecure.NewCredentials()))
	if err != nil {
		log.Fatalf("net.Connect connect: %v", err)
	}
	defer conn.Close()
	// 建立gRpc連線
	ctx := context.Background()
	grpcClient = pb.NewSimpleClient(conn)
	// 修改超時
	route(ctx, 2)

}

三、服務端判斷請求是否超時

當請求超時後,服務端應該停止正在進行的操作,避免資源浪費。

// Route 實現Route方法
func (s *SimpleService) Route(ctx context.Context, req *pb.SimpleRequest) (*pb.SimpleResponse, error) {
	data := make(chan *pb.SimpleResponse, 1)
	go handle(ctx, req, data)
	select {
	case res := <-data:
		return res, nil
	case <-ctx.Done():
		return nil, status.Errorf(codes.Canceled, "Client cancelled, abandoning.")
	}
}

func handle(ctx context.Context, req *pb.SimpleRequest, data chan<- *pb.SimpleResponse) {
	select {
	case <-ctx.Done():
		log.Println(ctx.Err())
		runtime.Goexit() //超時後退出該Go協程
	case <-time.After(4 * time.Second): // 模擬耗時操作
		res := pb.SimpleResponse{
			Code:  200,
			Value: "hello " + req.Data,
		}
		// //修改資料庫前進行超時判斷
		// if ctx.Err() == context.Canceled{
		// 	...
		// 	//如果已經超時,則退出
		// }
		data <- &res
	}
}

一般地,在寫庫前進行超時檢測,發現超時就停止工作。

完整server.go程式碼

package main

import (
	"context"
	pb "grpc/06deadlines/proto"
	"log"
	"net"
	"runtime"
	"time"

	"google.golang.org/grpc/codes"
	"google.golang.org/grpc/status"

	"google.golang.org/grpc"
)

/*
@author RandySun
@create 2022-03-27-20:03
*/

// SimpleService 定義我們的服務
type SimpleService struct {
}

// Route 實現Route方法
func (s *SimpleService) Route(ctx context.Context, req *pb.SimpleRequest) (*pb.SimpleResponse, error) {
	data := make(chan *pb.SimpleResponse)

	go handle(ctx, req, data)
	select {
	case res := <-data:
		return res, nil
	case <-ctx.Done():
		return nil, status.Errorf(codes.Canceled, "Client cancelled abandoning")

	}
}

// 超時處理
func handle(ctx context.Context, req *pb.SimpleRequest, data chan<- *pb.SimpleResponse) {

	select {
	case <-ctx.Done():
		log.Println(ctx.Err())
		runtime.Goexit() // 超時後退出該Go協程
	case <-time.After(4 * time.Second): // 模擬耗時操作
		res := pb.SimpleResponse{
			Code:  200,
			Value: "hello " + req.Data,
		}
		// //修改資料庫前進行超時判斷
		// if ctx.Err() == context.Canceled{
		// 	...
		// 	//如果已經超時,則退出
		// }
		data <- &res

	}
}

const (
	// Address 監聽地址
	Address string = ":8001"
	// NetWork 網路通訊協議
	NetWork string = "tcp"
)

func main() {
	// 監聽本地埠
	listener, err := net.Listen(NetWork, Address)
	if err != nil {
		log.Fatalf("net.Listen err: %V", err)
	}
	log.Println(Address, "net.Listing...")
	// 建立grpc服務例項
	grpcServer := grpc.NewServer()
	// 在grpc伺服器註冊我們的服務
	pb.RegisterSimpleServer(grpcServer, &SimpleService{})
	err = grpcServer.Serve(listener)
	if err != nil {
		log.Fatalf("grpcService.Serve err:%v", err)
	}
	log.Println("grpcService.Serve run succ")
}

四、執行結果

服務端:

go-grpc-example> cd .\06deadlines\server\
go-grpc-example\06deadlines\server> go build .\service.go
go-grpc-example\06deadlines\server> .\server.exe
2022/05/08 16:40:11 :8001 net.Listing...
2022/05/08 16:40:20 context deadline exceeded
2022/05/08 16:40:30 context deadline exceeded

客戶端:

go-grpc-example\06deadlines\client>  go build .\client.go
go-grpc-example\06deadlines\client> .\client.exe
2022/05/08 16:40:30 服務的返回響應data: <nil>
2022/05/08 16:40:30 Route timeout!

五、總結

超時時間的長短需要根據自身服務而定,例如返回一個hello grpc,可能只需要幾十毫秒,然而處理大量資料的同步操作則可能要很長時間。需要考慮多方面因素來決定這個超時時間,例如系統間端到端的延時,哪些RPC是序列的,哪些是可以並行的等等。
參考:https://grpc.io/blog/deadlines/