精品乱码久久久久久夜夜嗨,久久久久久亚洲Av无码精品专口,久久99国产精品二区不卡

net.LookupSRV()查询k8s无头服务

金庆 — Mon, 22 Feb 2021 08:17:00 GMT

net.LookupSRV()查询k8s无头服务

(金庆的专�?2021.2)

如下创徏 StatefulSet �?Headless Service: test.yaml

```
apiVersion: v1
kind: Service
metadata:
name: headless-svc
labels:
    app: headless-svc
spec:
ports:
- port: 80
    name: aaaa
- port: 20080
    name: bbbb
selector:
    app: headless-pod
clusterIP: None

---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: statefulset-test
spec:
serviceName: headless-svc
replicas: 3
selector:
    matchLabels:
      app: headless-pod
template:
    metadata:
      labels:
        app: headless-pod
    spec:
      containers:
      - name: myhttpd
        image: httpd
        ports:
        - containerPort: 80
        - containerPort: 20080
```

部��v�Q?br />```
kubectl apply -f test.yaml
```

然后开一�?shell:
```
kubectl run mygolang -it --image golang -- bash
If you don't see a command prompt, try pressing enter.
root@mygolang:/go#
```

先装�?vim:
```
apt install vim
```

然后写个 golang ��试�E�序�Q�向 DNS 查询 bbbb 服务的地址与端口：(x��)
```
root@mygolang:/jinqing# cat main.go
package main

import (
    "fmt"
    "github.com/davecgh/go-spew/spew"
    "net"
)

func main() {
    cname, addresses, err := net.LookupSRV("bbbb", "tcp", "headless-svc.default.svc.cluster.local")
    if err != nil {
        fmt.Printf("failed: %s\n", err)
    }
    fmt.Printf("cname: %s\n", cname)
    spew.Dump(addresses)
}
root@mygolang:/jinqing#
```

�q�行�l�果为：(x��)
```
cname: _bbbb._tcp.headless-svc.default.svc.cluster.local.
([]*net.SRV) (len=3 cap=4) {
(*net.SRV)(0xc00000e220)({
Target: (string) (len=58) "statefulset-test-0.headless-svc.default.svc.cluster.local.",
Port: (uint16) 20080,
Priority: (uint16) 0,
Weight: (uint16) 33
}),
(*net.SRV)(0xc00000e1e0)({
Target: (string) (len=58) "statefulset-test-2.headless-svc.default.svc.cluster.local.",
Port: (uint16) 20080,
Priority: (uint16) 0,
Weight: (uint16) 33
}),
(*net.SRV)(0xc00000e200)({
Target: (string) (len=58) "statefulset-test-1.headless-svc.default.svc.cluster.local.",
Port: (uint16) 20080,
Priority: (uint16) 0,
Weight: (uint16) 33
})
}
```

多次�q�行发现�l�果��Ҏ(gu��)��序固定，�q�没有按权重随机�?br />
Service yaml 定义中，必须为每个端口命名，不然没法查询�?br />
```
    cname, addresses, err := net.LookupSRV("bbbb", "tcp", "headless-svc.default.svc.cluster.local")
```
可利用默认域名后�~��写�ؓ(f��)
```
    cname, addresses, err := net.LookupSRV("bbbb", "tcp", "headless-svc")
```
输出相同�?br />

金庆 2021-02-22 16:17 发表评论

rpc应答太快造成��h��时

金庆 — Thu, 17 Sep 2020 07:59:00 GMT

rpc应答太快造成��h��时

(金庆的专�?2020.9)

在压��中发现��L��几个��h��时�Q�超时时长设大也�?x��)有�Q�而成功的��h��延时�q�小于超时时间�?br />查错的第一方向是查�|�络库中有消息丢失�?br />跟踪所有消息，发现��时的消息应该是正常处理�q�返回了�?br />于是查接收应�{�消息后的处理，最�l�找��C��码：(x��)
```go
func (c *Client) onRpcRet(cbIndex uint32, ...) {
   ii, ok := c.callbacks.Load(cbIndex)
   if !ok {
        // logger.Errorf("onRpcRet can not find cbIndex %d", cbIndex) // 可能是超时已�?br />        return
   }
```
打开日志�Q�发现超时的��h��对应有该条错误日志�?br />此处回调不存在的情况�Q�正常是先超时删除回调，然后再收到应�{��?br />现在是先收到了应�{�，发现找不到回调，然后�q�了一�D�|��间会(x��)被判��时无响应�?br />
��下面代码换个次序就好了�Q?br />```go
   if err := c.Session.Send(msg); err != nil {
        ...
        return
   }
   c.callbacks.Store(cbIndex, ...)
```
改�ؓ(f��)
```go
   // 必须先设回调�Q�然后发送，因�ؓ(f��)应答可能�?x��)很�?br />   c.callbacks.Store(cbIndex, ...)
   if err := c.Session.Send(msg); err...
```

压测时因为加压机CPU是满负蝲�q��{�Q�所�?Send() �?Store() 之间可能�?x��)间隔数毫秒�Q?br />��_�� rpc ��h��处理完成�q�返回，而应�{�返回时回调�q�没讄��?br />
�?Send() �?Store() 写代码会(x��)�E�微��单点�Q�因�?Send() ��p�|后可以直接返回�?br />�?Store() �?Send() �Ӟ��Send() ��p�|则需要相�?Delete().

金庆 2020-09-17 15:59 发表评论

金庆 — Fri, 11 Sep 2020 06:55:00 GMT

# golang各数值类型的最大最��?br />
(金庆的专�?2020.9)

golang �?math 包已�l�定义了以下帔R��Q?br />
Constants

```
const (
    E   = 2.71828182845904523536028747135266249775724709369995957496696763 // https://oeis.org/A001113
    Pi = 3.14159265358979323846264338327950288419716939937510582097494459 // https://oeis.org/A000796
    Phi = 1.61803398874989484820458683436563811772030917980576286213544862 // https://oeis.org/A001622

    Sqrt2   = 1.41421356237309504880168872420969807856967187537694807317667974 // https://oeis.org/A002193
    SqrtE   = 1.64872127070012814684865078781416357165377610071014801157507931 // https://oeis.org/A019774
    SqrtPi = 1.77245385090551602729816748334114518279754945612238712821380779 // https://oeis.org/A002161
    SqrtPhi = 1.27201964951406896425242246173749149171560804184009624861664038 // https://oeis.org/A139339

    Ln2    = 0.693147180559945309417232121458176568075500134360255254120680009 // https://oeis.org/A002162
    Log2E = 1 / Ln2
    Ln10   = 2.30258509299404568401799145468436420760110148862877297603332790 // https://oeis.org/A002392
    Log10E = 1 / Ln10
)
```

Mathematical constants.

```
const (
    MaxFloat32             = 3.40282346638528859811704183484516925440e+38 // 2**127 * (2**24 - 1) / 2**23
    SmallestNonzeroFloat32 = 1.401298464324817070923729583289916131280e-45 // 1 / 2**(127 - 1 + 23)

    MaxFloat64             = 1.797693134862315708145274237317043567981e+308 // 2**1023 * (2**53 - 1) / 2**52
    SmallestNonzeroFloat64 = 4.940656458412465441765687928682213723651e-324 // 1 / 2**(1023 - 1 + 52)
)
```

Floating-point limit values. Max is the largest finite value representable by the type. SmallestNonzero is the smallest positive, non-zero value representable by the type.

```
const (
    MaxInt8   = 1<<7 - 1
    MinInt8   = -1 << 7
    MaxInt16 = 1<<15 - 1
    MinInt16 = -1 << 15
    MaxInt32 = 1<<31 - 1
    MinInt32 = -1 << 31
    MaxInt64 = 1<<63 - 1
    MinInt64 = -1 << 63
    MaxUint8 = 1<<8 - 1
    MaxUint16 = 1<<16 - 1
    MaxUint32 = 1<<32 - 1
    MaxUint64 = 1<<64 - 1
)
```
Integer limit values.

金庆 2020-09-11 14:55 发表评论

金庆 — Fri, 24 Jul 2020 01:15:00 GMT

参数太灵�z�d��易出�?br />
(金庆的专�?2020.7)

golang中可以将参数�c�d��设�ؓ(f��) interface{}, �q�样��可以传入�Q意类型的参数�Q?br />�?C++ �?void* 的作用相伹{�?br />但是�q�种万能�c�d��应该��量��用�Q�尽量��用具体的�c�d��Q�或者��用一个具体的接口�c�d��?br />主要的原因是, 让编译期的类型检查挡住编码错误，减少�q�行期的错误�?br />
例如�Q�go-mongo-driver 有个创徏索引的参敎ͼ�(x��)
```
type IndexModel struct {
    // A document describing which keys should be used for the index. It cannot be nil.
    // This must be an order-preserving type such as bson.D. Map types such as bson.M are not valid.
    Keys interface{}
    ...
}
```

其中 Keys 可以是�Q意类型，�?1234, "abcd", 当然不符合烦引要求的�c�d��?x��)返回失败�?br />但是 bson.M �c�d��Q�会(x��)创徏索引成功�Q�但是烦引的�ơ序�?x��)有错误�?br />注释中已指出�Q�不要用 bson.M, 应该使用 bson.D.

正确�?Keys 如下�Q�表�C�复合烦�?(field1, field2)�Q?表示正序�Q?1则反序：(x��)
```
    indexModel.Keys := bson.D{{"field1", 1}, {"field2", 1}}
```

如果使用 bson.M, 实际上是�?map�Q?br />```
    indexModel.Keys := bson.M{"field1"�Q?, "field2": 1}
```
因�ؓ(f��) map 成员的次序不定，最后创建的索引可能�?(field1, field2)�Q�也可能�?(field2, field1)�?br />
此处�c�d��允许 interface{} 的想法是�Q�允�怓Q意类型，�?bson �~�码后传�l?mongo 服务器，�q�不�?x��)进行类型检查�?br />�q�种灉|��性非常容易造成错误�Q��ƈ且如何��用也不明��? 仅靠注释作用很小�?br />

金庆 2020-07-24 09:15 发表评论

金庆 — Fri, 17 Apr 2020 02:08:00 GMT

# grpc外部负蝲均衡器测�?br />
(金庆的专�?2020.4)

grpc �Ҏ(gu��)��个请求进行负载均衡。负载均衡的方式有：(x��)

* 代理模式
* 客户端实�?br />* 外部负蝲均衡

参考：(x��)gRPC LB https://blog.csdn.net/xiaojia1100/article/details/78842295

gRPC 中负载均衡的主要机制是外部负载均衡�?br />
gRPC 定义了外部负载均衡服务的接口�Q�https://github.com/grpc/grpc/tree/master/src/proto/grpc/lb/v1

* load_balancer.proto 客户端向 lb 服查询后端列�?br />* load_reporter.proto lb 服向后端服查询负�?br />
https://github.com/bsm/grpclb 实现了一�?grpc 的外部负载均衡服�?br />因�ؓ(f��)其实现早于负载均衡服的接口规范，所以接口定义与 grpc 规范不同�?br />�?issue#26: https://github.com/bsm/grpclb/issues/26#issuecomment-613873655
grpclb 目前仅支�?consul 服务发现�?br />
标准�?grpclb 实现目前好像只有 https://github.com/joa/jawlb�?br />jawlb 通过 Kubernetes API 来发现服务�?br />
以下��试 grpc 客户端从 jawlb 服查询服务器列表�Q�然后请求服务�?br />首先在本机开了多�?greeter 服实例，端口不同�?br />然后更改 greeter 客户端，不要直接�q?greeter 服地址�Q�而是配一�?jawlb 服地址�?br />同时更改 jawlb, 删除服务发现�Q�改为固定输出本机服务列表，定时切换�?br />
greeter 是指 grpc-go 中的例子�Q�grpc-go\examples\helloworld\greeter

## greeter 服更�?br />
��d��参数指定服务端口�?br />
```
package main

import (
    "fmt"
    "log"
    "net"

    "github.com/spf13/pflag"
    "github.com/spf13/viper"
    "golang.org/x/net/context"
    "google.golang.org/grpc"
    pb "google.golang.org/grpc/examples/helloworld/helloworld"
)

// GreeterServer is used to implement helloworld.GreeterServer.
type GreeterServer struct {
}

// SayHello implements helloworld.GreeterServer
func (s *GreeterServer) SayHello(ctx context.Context, in *pb.HelloRequest) (*pb.HelloReply, error) {
    msg := fmt.Sprintf("Hello %s from server-%d", in.Name, viper.GetInt("port"))
    return &pb.HelloReply{Message: msg}, nil
}

func main() {
    pflag.Int("port", 8000, "server bind port")
    pflag.Parse()
    viper.BindPFlags(pflag.CommandLine)
    port := viper.GetInt("port")

    addr := fmt.Sprintf(":%d", port)
    lis, err := net.Listen("tcp", addr)
    if err != nil {
        log.Fatalf("failed to listen: %v", err)
    }
    s := grpc.NewServer()
    pb.RegisterGreeterServer(s, &GreeterServer{})
    s.Serve(lis)
}
```

## greeter 客户端更�?br />```
package main

import (
    "context"
    "log"
    "os"
    "time"

    "github.com/sirupsen/logrus"
    "google.golang.org/grpc"
    _ "google.golang.org/grpc/balancer/grpclb"
    pb "google.golang.org/grpc/examples/helloworld/helloworld"
    "google.golang.org/grpc/grpclog"
    "google.golang.org/grpc/resolver"
    "google.golang.org/grpc/resolver/manual"
)

const (
    defaultName = "world"
)

func init() {
    grpclog.SetLogger(logrus.New())
}

func main() {
    rb := manual.NewBuilderWithScheme("whatever")
    rb.InitialState(resolver.State{Addresses: []resolver.Address{
        {Addr: "127.0.0.1:8888", Type: resolver.GRPCLB},
    }})

    conn, err := grpc.Dial("whatever:///this-gets-overwritten", grpc.WithInsecure(), grpc.WithBlock(),
        grpc.WithResolvers(rb))
    if err != nil {
        log.Fatalf("did not connect: %v", err)
    }
    defer conn.Close()
    c := pb.NewGreeterClient(conn)

    name := defaultName
    if len(os.Args) > 1 {
        name = os.Args[1]
    }

    for {
        ctx, cancel := context.WithTimeout(context.Background(), 3*time.Second)
        r, err := c.SayHello(ctx, &pb.HelloRequest{Name: name})
        cancel()

        if err != nil {
            log.Fatalf("could not greet: %v", err)
            time.Sleep(time.Second)
            continue
        }

        log.Printf("Greeting: %s", r.GetMessage())
        time.Sleep(time.Second)
    }
}
```

有以下更改：(x��)
* import _ "google.golang.org/grpc/balancer/grpclb"
* grpc.Dial("whatever:///this-gets-overwritten", grpc.WithResolvers(rb))
    + 采用一个自定义解析器，用来获取 jawlb 地址
    + Scheme("whatever") 可以��L��Q�用作解析器名字
    + 目标 this-gets-overwritten 可以��L��Q�因�?jawlb 忽略了该名字
    + 127.0.0.1:8888 �?jawlb 地址
* 改�ؓ(f��)每秒��h��一��?br />
正常�?grpclb 是在 DNS 中设�|?SRV 记录�Q?br />此处��试避免讄�� DNS, 采用了一个自定义解析器，
代码上多了几行�?br />�?DNS 讄��的好处是, 可以直接解析为后�?IP, 也可以添�?grpclb, 代码上如同直接连接后端：(x��)
```
    conn, err := grpc.Dial("dns:///myservice.domain.com", grpc.WithInsecure())
```

## jawlb 更改

### main.go

删除所有配�|�，改�ؓ(f��)固定本机 8888 端口监听�?br />
* 删除 `envconfig.MustProcess("JAWLB", &cfg)`
* listen() 改�ؓ(f��)
    ```
    func listen() (conn net.Listener, err error) {
        conn, err = net.Listen("tcp", ":8888")
        return
    }
    ```

### watch.go

```
package main

import (
    "context"
    "fmt"
    "net"
    "time"
)

func watchService(ctx context.Context) (_ <-chan ServerList, err error) {
    ch := make(chan ServerList)

    go func() {
        ticker := time.NewTicker(10 * time.Second)
        i := 0
        for {
            select {
            case <-ctx.Done():
                ticker.Stop()
                close(ch)
                return
            case <-ticker.C:
                i += 1
                fmt.Printf("i = %d\n", i)
                ports := []int32{8010, 8020}
                var servers []Server
                for _, port := range ports {
                    servers = append(servers, Server{IP: net.ParseIP("127.0.0.1"), Port: port + int32(i%2)})
                }
                ch <- servers
            } // select
        } // for
    }()

    return ch, nil
}
```

删除所有服务发��C��码，改�ؓ(f��)�?0�U�切换端口：(x��)8010,8020 <-> 8011,8021

## �q�行

### jawlb
```
λ jawlb.exe
2020/04/16 15:35:17 waiting for TERM
i = 1
2020/04/16 15:35:27 endpoints:
2020/04/16 15:35:27     127.0.0.1:8011
2020/04/16 15:35:27     127.0.0.1:8021
i = 2
2020/04/16 15:35:37 endpoints:
2020/04/16 15:35:37     127.0.0.1:8010
2020/04/16 15:35:37     127.0.0.1:8020
```

### server

�q�行 4 个实例：(x��)
```
server --port 8010
server --port 8020
server --port 8011
server --port 8021
```

### client

```
λ client
INFO[0002] lbBalancer: handle SubConn state change: 0xc00008a590, CONNECTING
INFO[0002] Channel Connectivity change to CONNECTING
INFO[0002] lbBalancer: handle SubConn state change: 0xc00008a5f0, CONNECTING
INFO[0002] Subchannel picks a new address "127.0.0.1:8021" to connect
INFO[0002] Subchannel Connectivity change to READY
INFO[0002] lbBalancer: handle SubConn state change: 0xc00008a590, READY
INFO[0002] Channel Connectivity change to READY
INFO[0002] Subchannel Connectivity change to READY
INFO[0002] lbBalancer: handle SubConn state change: 0xc00008a5f0, READY
2020/04/16 15:37:47 Greeting: Hello world from server-8021
2020/04/16 15:37:48 Greeting: Hello world from server-8011
2020/04/16 15:37:49 Greeting: Hello world from server-8021
2020/04/16 15:37:50 Greeting: Hello world from server-8011
2020/04/16 15:37:51 Greeting: Hello world from server-8021
2020/04/16 15:37:52 Greeting: Hello world from server-8011
2020/04/16 15:37:53 Greeting: Hello world from server-8021
2020/04/16 15:37:54 Greeting: Hello world from server-8011
2020/04/16 15:37:55 Greeting: Hello world from server-8021
2020/04/16 15:37:56 Greeting: Hello world from server-8011
INFO[0012] lbBalancer: processing server list: servers: servers:
INFO[0012] lbBalancer: server list entry[0]: ipStr:|127.0.0.1|, port:|8020|, load balancer token:||
INFO[0012] lbBalancer: server list entry[1]: ipStr:|127.0.0.1|, port:|8010|, load balancer token:||
2020/04/16 15:37:57 Greeting: Hello world from server-8020
2020/04/16 15:37:58 Greeting: Hello world from server-8010
2020/04/16 15:37:59 Greeting: Hello world from server-8020
2020/04/16 15:38:00 Greeting: Hello world from server-8010
2020/04/16 15:38:01 Greeting: Hello world from server-8020
2020/04/16 15:38:02 Greeting: Hello world from server-8010
2020/04/16 15:38:03 Greeting: Hello world from server-8020
2020/04/16 15:38:04 Greeting: Hello world from server-8010
2020/04/16 15:38:05 Greeting: Hello world from server-8020
2020/04/16 15:38:06 Greeting: Hello world from server-8010
INFO[0022] lbBalancer: processing server list: servers: servers:
INFO[0022] lbBalancer: server list entry[0]: ipStr:|127.0.0.1|, port:|8021|, load balancer token:||
INFO[0022] lbBalancer: server list entry[1]: ipStr:|127.0.0.1|, port:|8011|, load balancer token:||
2020/04/16 15:38:07 Greeting: Hello world from server-8011
2020/04/16 15:38:08 Greeting: Hello world from server-8021
2020/04/16 15:38:09 Greeting: Hello world from server-8011
```

## �l�论

客户端应用一个自定义 resolver 解析 "whatever:///this-gets-overwritten"�Q?br />获取�?`{Addr: "127.0.0.1:8888", Type: resolver.GRPCLB}`,
知道�q�是一�?grpclb�Q�于是按 load_balancer.proto 的定义查�?jawlb 来获取后端地址列表�?br />
jawlb �?10s 更新一�ơ服务器列表�Q�每�ơ输出多个地址。客��L(f��ng)��在多个地址间轮换请求�?br />
## 其他��试

* 不开 jawlb�Q�客��L(f��ng)��无法成功请求，直到 jawlb 开启才成功
* 中途关�?jawlb, ��h��仍会(x��)成功�Q�但是保持�ؓ(f��)最后的服务器列�?br />    + 同时�?x��)不断尝试重�q?jawlb, 但是重连成功后没有切换服务，应该是个错误
* Dial() 不加 grpc.WithBlock() 参数, 报错�Q�all SubConns are in TransientFailure

```
λ client
INFO[0000] parsed scheme: "whatever"
INFO[0000] ccResolverWrapper: sending update to cc: {[{127.0.0.1:8888 1 }] }
INFO[0000] ClientConn switching balancer to "grpclb"
INFO[0000] Channel switches to new LB policy "grpclb"
INFO[0000] lbBalancer: UpdateClientConnState: {ResolverState:{Addresses:[{Addr:127.0.0.1:8888 ServerName: Attributes: Type:1 Metadata:}] ServiceConfig: Attributes:} BalancerConfig:}
INFO[0000] parsed scheme: "grpclb-internal"
INFO[0000] ccResolverWrapper: sending update to cc: {[{127.0.0.1:8888 0 }] }
INFO[0000] ClientConn switching balancer to "pick_first"
INFO[0000] Channel switches to new LB policy "pick_first"
INFO[0000] Subchannel Connectivity change to CONNECTING
INFO[0000] blockingPicker: the picked transport is not ready, loop back to repick
INFO[0000] pickfirstBalancer: HandleSubConnStateChange: 0xc00003fb10, {CONNECTING }
INFO[0000] Channel Connectivity change to CONNECTING
INFO[0000] Subchannel picks a new address "127.0.0.1:8888" to connect
INFO[0000] CPU time info is unavailable on non-linux or appengine environment.
INFO[0000] Subchannel Connectivity change to READY
INFO[0000] pickfirstBalancer: HandleSubConnStateChange: 0xc00003fb10, {READY }
INFO[0000] Channel Connectivity change to READY
INFO[0000] lbBalancer: processing server list: servers: servers:
INFO[0000] lbBalancer: server list entry[0]: ipStr:|127.0.0.1|, port:|8010|, load balancer token:||
INFO[0000] lbBalancer: server list entry[1]: ipStr:|127.0.0.1|, port:|8020|, load balancer token:||
INFO[0000] Subchannel Connectivity change to CONNECTING
INFO[0000] Subchannel Connectivity change to CONNECTING
INFO[0000] Channel Connectivity change to TRANSIENT_FAILURE
INFO[0000] lbBalancer: handle SubConn state change: 0xc00008a220, CONNECTING
INFO[0000] Channel Connectivity change to CONNECTING
INFO[0000] lbBalancer: handle SubConn state change: 0xc00008a280, CONNECTING
2020/04/16 16:40:06 could not greet: rpc error: code = Unavailable desc = all SubConns are in TransientFailure
```

金庆 2020-04-17 10:08 发表评论

数组make参数错误

金庆 — Thu, 19 Mar 2020 05:38:00 GMT

# 数组make参数错误

(金庆的专�?2020.3)

�?mgo �?Bulk 接口操作 mongodb 的代码，�?mongodb 4.2 正常�Q?br />而连 mongodb 2.6 报错�Q?br />```
error parsing element 0 of field documents :: caused by :: wrong type for '0' field, expected object, found 0: null
```

代码如下�Q?br />```
func (u *Util) AddUsers(userIDs []int) error {
    docs := make([]interface{}, len(userIDs))
    for _, userID := range userIDs {
        docs = append(docs, dbdoc.UsersDoc{
            UserID:     userID,
        })
    }
    bulk := u.c().Bulk()
    bulk.Insert(docs...)
    _, err := bulk.Run()
    return err
}
```

试着�Ҏ(gu��)��如下调用��成功：(x��)
```
func (u *Util) AddUsers(userIDs []int) error {
    bulk := u.c().Bulk()
    for _, userID := range userIDs {
        bulk.Insert(dbdoc.UsersDoc{
            UserID:     userID,
        })
    }
    _, err := bulk.Run()
    return err
}
```

�?Insert() 无论是多个还是单个参敎ͼ�底层实现是一��L(f��ng)��?br />
输入单个�?[]int{123} �q�是出错.
��?Insert(docs...) �Ҏ(gu��)�� Insert(docs[0]) 时发现元素�ؓ(f��)�I��?br />最�l�找��C��错误的代码行�Q?br />```
-    docs := make([]interface{}, len(userIDs))
+    docs := make([]interface{}, 0, len(userIDs))
```

高版本的mongodb�l�果正确应该是他忽略了空的文档�?br />
感觉 golang make() 卛_��?个参��C��可以3个参数调用的设计是个错误�?br />

金庆 2020-03-19 13:38 发表评论

go不要导出channel

金庆 — Tue, 10 Mar 2020 07:58:00 GMT

# go不要导出channel

(金庆的专�?2020.3)

导出 channel 的包都比较难用。��^�怹�不会(x��)��接口设计成 channel.
以下摘自�Q�https://studygolang.com/articles/12135?fr=sidebar

不要导出�q�发原语

Go 提供了非常易于��用的�q�发原语�Q�这也导致了它被�q�度的��用。我们主要担心的�?channel �?sync package 。有的时候我们会(x��)导出一�?channel �l�用户��用。另外一个常见的错误��是在��?sync.Mutex 作�ؓ(f��)�l�构体字�D늚�时候没有把它设�|�成�U�有。这�q�不��L��很糟�p�，不过在写��试的时候却需要考虑的更加全面�?br />
当我们导�?channel 的时候我们就��个包的用户带来了��试上的一些不必要的麻烦。你每导��Z��?channel ��是在提高用户在��试时候的隑ֺ�。�ؓ(f��)了写出正��的��试�Q�用户必��考虑�q�些�Q?br />
什么时候数据发送完成�?br /> 在接受数据的时候是否会(x��)发生错误�?br /> 如果需要在包中清理使用�q�的channel的时候该怎么做�?br /> 如何��?API ��装成一个接口，使我们不用直接去调用它�?br />

金庆 2020-03-10 15:58 发表评论

go代码覆盖��试

金庆 — Sat, 29 Feb 2020 01:10:00 GMT

# go代码覆盖��试

(金庆的专�?2020.2)

假设应用名�ؓ(f��) MyMod, ��d��数的文�g�?main.go, 那就创徏如下�?main_test.go 文�g�Q?br />```
package main

/* ��试整个服务器�?/span>
    go test -c -covermode=count -coverpkg ./...
生成 MyMod.test.exe. 复制�?bin 目录下运行：(x��)
    MyMod.test.exe --systemTest --test.coverprofile MyMod.cov
生成代码覆盖��试�l�果 MyMod.cov, 需要在 go.mod ��理的目录下执行
    go tool cover -html=MyMod.cov -o MyMod.html
打开 MyMod.html 查看�l�果�?/span>
*/

import (
    "flag"
    "fmt"
    "testing"
)

var systemTest *bool

func init() {
    systemTest = flag.Bool("systemTest", false, "Set to true when running system tests")
}

// Test started when the test binary is started. Only calls main.
func TestSystem(t *testing.T) {
    if *systemTest {
        fmt.Println("Test system...")
        main()
    }
}
```

MyMod.cov �?x��)积累，可加�?Git.
MyMod.cov 提交前，需对其排序�Q�方便查�?diff.

排序脚本 sort_cov.bat 如下�Q?br />```
head -1 MyMod.cov > tmp.cov
cat MyMod.cov | sed '1d' | sort >> tmp.cov
cp tmp.cov MyMod.cov
del tmp.cov
pause
```

金庆 2020-02-29 09:10 发表评论

金庆 — Sat, 22 Feb 2020 04:45:00 GMT

# 考察go一致性hash�?/h1>
(金庆的专�?2020.2)

github �?consistent, 按星��C��ơ察�?br />

## stathat/consistent

Consistent hash package for Go.

688�?

stathat 应该是个数据�l�计的云服务。consistent被认为是生��可用的�?br />所有文档都�?godoc. �C�Z��和接口简�z�易懂�?br />

## lafikl/consistent

A Go library that implements Consistent Hashing and Consistent Hashing With Bounded Loads.

575 �?br />
有界负蝲的一致性哈希算�?br />
�C�Z��中是 c.GetLeast() 获取最��负载，不像是一致性hash的应用�?br />从代码看�Q�应该是最�q�的未满载，不是最��负载�?br />
`MaxLoad()`说明昄��最大负载是自动讄��?(total_load/number_of_hosts)*1.25�?br />
`const replicationFactor = 10`

没有一�ơ获取多个的接口�?br />

## serialx/hashring

Consistent hashing "hashring" implementation in golang (using the same algorithm as libketama)

367 �?br />
是从 Python 库移植的。可以设�|?weight.

�C�Z��
```
replicaCount := 3
ring := hashring.New(serversInRing)
server, _ := ring.GetNodes("my_key", replicaCount)
```
实际上是 GetN 的功能。内部实现缺��?replicationFactor

## buraksezer/consistent

Consistent hashing with bounded loads in Golang

263 �?br />
未提供缺省hash函数�?br />
多了一个`PartitionCount`

c.loads 仅用于查询负载，�q�没有用于有界负载�?br />
�l�论是这个不是有界负载的一致性哈希，同时一致性哈希的实现不同��d��?br />

## 其他

jump hash 不能�U�除节点�Q�不考虑

### dgryski/go-jump

go-jump: Jump consistent hashing

281 �?br />

### lithammer/go-jump-consistent-hash

⚡️ Fast, minimal memory, consistent hash algorithm

138 �?img src ="http://www.shnenglu.com/jinq0123/aggbug/217148.html" width = "1" height = "1" />

金庆 2020-02-22 12:45 发表评论

��go函数指针转�ؓ(f��)接口

金庆 — Thu, 13 Feb 2020 06:38:00 GMT

# ��go函数指针转�ؓ(f��)接口

(金庆的专�?2020.2)

golang 中的接口如下�Q?br />
```
type Writer interface {
    Write func(p []byte) (n int, err error)
}
```

一般API参数要求一个接口，而不是一个函数指针，�?io.Copy() 需要输入一�?Writer �?Reader�Q?br />```
func Copy(dst Writer, src Reader) (written int64, err error)
```

而不是这�?个函数指针：(x��)
```
func CopyWithFunc(writeFunc func([]byte) (int, error), readRunc func([]byte) (int, error)) (written int64, err error)
```

大家�l�一使用接口�Q�而不是接口和函数指针��L(f��ng)��Q�可以避免API复杂化�?br />�?io.Copy() �?个参敎ͼ�如果要支持接口和函数指针��L(f��ng)��Q�就�?x��)变�?�?Copy() 重蝲�?br />golang 没有重蝲�Q�就只能�?个不同的函数名�?br />
在实际��用中�Q�需要将函数转化成接口，才能调用 io.Copy().
如有一个函�?
```
func MyWriteFunction(p []byte) (n int, err error) {
    fmt.Print("%v",p)
    return len(p),nil
}
```
调用 io.Copy() 旉��要创��Z��?Writer�Q��ƈ��该函数指针转型为Writer后��用�?br />�q�里�?`WriteFunc` �c�d��实现 Writer�?br />
```
type WriteFunc func(p []byte) (n int, err error)

func (wf WriteFunc) Write(p []byte) (n int, err error) {
    return wf(p)
}
```

WriteFunc 本��n是个�?MyWriteFunction 同类型的函数�c�d��Q�同时实��C�� Writer 接口�?br />所�?MyWriteFunction 可以直接转成WriteFunc�c�d��成�ؓ(f��)一�?Writer.
�q�样��可以调�?io.Copy() 了：(x��)

```
io.Copy(WriteFunc(MyWriteFunction), strings.NewReader("Hello world"))
```

参考：(x��)https://stackoverflow.com/questions/20728965/golang-function-pointer-as-a-part-of-a-struct

金庆 2020-02-13 14:38 发表评论

golang chan 关闭时的原则

金庆 — Wed, 08 Jan 2020 11:46:00 GMT

golang chan 关闭时的原则

(金庆的专�?2020.1)

golang �E�序中检��到 DATA RACE, �?chan 关闭和发送冲�H�：(x��)

==================
WARNING: DATA RACE
Write at 0x00c000098010 by goroutine 68:
runtime.closechan()
      /usr/lib/golang/src/runtime/chan.go:327 +0x0
valky/common/tcp.(*Session).Close()
      /var/tmp/src/f4f4f712-7894-4d98-83dd...
valky/common/tcp.(*Session).recvloop()
      /var/tmp/src/f4f4f712-7894-4d98-83dd...

Previous read at 0x00c000098010 by goroutine 100:
runtime.chansend()
      /usr/lib/golang/src/runtime/chan.go:140 +0x0
valky/common/tcp.(*Session).Send()
      /var/tmp/src/f4f4f712-7894-4d98-83dd...
main.(*Role).sendMsg()
      /var/tmp/src/f4f4f712-7894-4d98-83dd...
==================
Found 1 data race(s)

查了一�?chan 关闭的正��做法，发现了一��非常详�l�的文章�Q?br />[How to Gracefully Close Channels](https://go101.org/article/channel-closing.html)

文中指出�Q�chan 多次关闭�Q�或者在关闭�?chan 上发送，都会(x��) panic.
上面�?DATA RACE 属于�q�运�Q�没�?panic�?br />
chan 关闭时的原则是：(x��)不要在接收协�E�中关闭�Q��ƈ且，如果有多个发送者时��׃��要关闭chan了�?br />
上面的DATA RACE 是在接收协程中关闭chan.

文中详细列出了多�U�方案关闭chan.
如果�_�暴点，可以直接加个 recover. 其他�Ҏ(gu��)��都是要保证发送完成后再关闭�?br />

金庆 2020-01-08 19:46 发表评论

open-match匚w��程

金庆 — Thu, 31 Jan 2019 02:21:00 GMT

# open-match匚w��程

(金庆的专�?2019.1)

https://github.com/GoogleCloudPlatform/open-match

open-match 是一个通用的游戏匹配框架�?br />由游戏提供自定义的匹配算法（以docker镜像的方式提供）�?br />
分�ؓ(f��)多个�q�程�Q�各�q�程之间�׃�n一�?redis.

* 前端, 接收玩家加入 redis�Q�成功后通知玩家戉K��服地址
* 后端�Q�设�|�一局游戏的匹配规则，讄��戉K��服地址
* MMFOrc�Q�启动匹配算�?MMF)
* MMF, 自定义匹配算法，��d�� redis 获取玩家�Q�匹配成功就��结果写�?redis. 仅匹配一局��退出�?br />
游戏服中�q�接 open-match 的前端与后端的进�E�，分别�U�Cؓ(f��) frontendclient �?Director�?br />输入�?部䆾�Q�一是玩家信息，二是对局信息�?br />Director 向后端输入对局信息�Q�就�?x��)收��C��个接一个的对局人员列表.
Director 需要�ؓ(f��)每个对局开戉K��Q�然后通知后端戉K��地址�?br />后端��房间地址写入 redis, 然后前端��d��到房间地址�Q�就通知 frontendclient�Q�让玩家�q�入戉K��?br />

## test/cmd/frontendclient

模拟大厅服或�l�队服，�q�接前端API, ��h��匚w��玩家/队伍。成功后��收到房间服(DGS)的地址(Assignment)�?br />
Player 实际上是一个队伍，其中ID字段是用�I�格分隔的多个ID.
虽然参数�c�d��都是 Player, CreatePlayer() 参数为整个队伍，�?GetUpdates() 参数是单个玩家�?br />
main() 中创建多个玩�Ӟ��每个玩家调用 GetUpdates() 以获取结果，go waitForResults() 中处理结果�?br />waitForResult() ��d��中的匹配结果，压入 resultsChan�Q�但好像 resultsChan 仅用于打华ͼ��?br />所有玩家合�q�到 g 实例中，然后调用 CreatePlayer() ��h��匚w��?br />
cleanup() 调用 DeletePlayer() 来删除匹配请求，不仅需删除整个队伍�Q�也需要删除单个玩家�?br />
好像最后取�l�果没取对地方，应该�?resultChan 中获�?Assignment, �q�用该地址 udpClient().

看了该示例就可以理解 frontend.proto

## examples/backendclient

MatchObject.Properties 是从 testprofile.json ��d��的，应该改名�?Profile 是否更好点？
pbProfile �?MatchObject�Q�Profile �{�同�?MatchObject?
Profile 的定义是 MMF 所需的所有参数�?br />`pbProfile.Properties = jsonProfile` 重复�?遍�?br />
ListMatches()列出�q�个Profile的所有匹配�?br />收到一个匹配后�Q�须用CreateAssignments()��房间服地址, �U�Cؓ(f��) Assignment, 发送到所有游戏客��L(f��ng)��?br />

## cmd/frontendapi

CreatePlayer() ��?Player 对象写入 redis, 键��gؓ(f��) Player.Id, �c�d��?HSET�?br />�?Player 的每�?attribute�Q�添加到 ZSET 中去�?br />此处 Player 是一�l�玩家�?br />
GetUpdates() 每隔2s��d��redis, Player数据有变化时��发送。此�?Player 是单个玩家�?br />
如果CreatePlayer()中队伍只有一个玩�Ӟ��
则写入的Player与GetUpdates()中读取的玩家是同一个redis键�?br />

## cmd/backendapi

CreateMatch() �?profile �c�d��?MatchObject, 是一个比赛的限制条�g�?br />profile 先写�?redis, 键�ؓ(f��) profile.Id.
`requestKey := xid() + "." + profile.Id`,
�q�将 requestKey 加入 redis 集合 "profileq"�?br />然后�?s查询 redis, 看是否有 requestKey 键出玎ͼ��q�返回该倹{�?br />
ListMatch() �?s调用一��?CreateMatch().

DeleteMatch() 仅仅删除 Id �q�个键�?br />
CreateAssignments() 为多个队伍设�|�Assignment, ��x��间地址�?br />遍历所有Roster中的Player对象�Q�在redis中设�|�Assignment.
(Assignment 更改后，�?x��)触发前端更新�?
��所�?Player.Id �?"proposed" �U�d�� "deindexed"�Q�这两个�?ZSET, 分��gؓ(f��)加入旉��?br />Roster 应该是比赛中的阵营，如红方，蓝方�Q�每个阵营中可有多个队伍�?br />
DeleteAssignments() 仅仅遍历所�?Player 对象来删�?Assignment 字段�?br />

## cmd/mmforc

匚w��程是由 mmforc (matchmaking function orchestrator) 控制的�?br />
mmforc 每秒�?redis �?profileq 中取�?100 个成�? 其中 profileq 是个set�c�d��Q?br />使用命��o(h��)为`SPOP profileq 100`.

�Ҏ(gu��)��?profile, 创徏一�?k8s ��d��Q?br />
```
// Kick off the job asynchrnously
go mmfunc(ctx, profile, cfg, defaultMmfImages, clientset, &pool)
```

每隔10s, �q�有所有匹配�Q务都完成后，需�?`checkProposals`, 卛_��?evaluator ��d��?br />
profileq 中的元素 profile 为字�W�串�Q�matchObjectID.profileID�?br />�?profileID 为键�Q�可以从 redis ��d�� profile 的内�? profile 是个 MatchObject 对象�?br />
profile 的内容�ؓ(f��) json �Ԍ��其中 "jsonkeys.mmfImages" �?mmf (matchmaking function) 镜像�?br />
如果profile��d��p�|�Q�或�?mmfImages 为空�Q�则使用默认的镜像。mmfImages 未来�?x��)支持多个镜像�?br />
通过 MMF_* 环境变量传入各种参数.

## mmf

�C�Z��Q�examples\functions\golang\manual-simple

从环境变�?"MMF_PROFILE_ID" 解析�?profileID, �q�向 redis 查询(HGETALL) profile�Q�HSET �c�d��?br />
�?profile 中取 pools 字段�Q�即匚w��条�g�?br />pools 分�ؓ(f��)多个 pool, 每个 pool 中有多个 filter, 每个 filter �?redis 取符合的 Player.

profile 用到以下字段�Q?br />
* "properties.playerPool"
json�Ԍ��是一些过滤条�Ӟ��?#8220;mmr: 100-999”
* "properties.roster"
json�? 是多个队伍大��，�?“red: 4”

�C�Z��见：(x��)`examples\backendclient\profiles\testprofile.json`

### ��单匹配过�E?/h3>
simple mmf 的匹配过�E�如下：(x��)

1. �?redis 查询 profile�Q�获取过滤条件和各队伍大��?br />1. 每个�q��o(h��)条�g�?redis 查询�Q�所有结果的交集为可选成�?br />1. 去除 ignoreList, ��x��q?800s 内已匚w��成功的成员，�?proposal �?deindexed ZSET 列表�?br />1. 如果可选成员个数太��，�?insufficient_players �q��?br />1. 分配各个队伍成员
1. �?redis 记录�l�果

### �l�果

profile 中添�?roster�Q�即各阵营成员名单，存入 prososalKey.
保存不分队伍的成员名单�?br />然后�?"proposalq" ��d�� prososalKey

### �l�节

poolRosters �?(pool�? filter attribute) 为键�Q��gؓ(f��) Player ID 列表.
保存�?redis 查询的符合条件的 Player ID.

overlaps �?pool 名�ؓ(f��)键，保存�W�合该pool中所有filter�?Player ID 列表�Q�去�?ignore list.

rosters �?profile 中的 "properties.rosters" 字段。不知何用？
遍历 rosters, 为每个阵营的每个player扑ֈ�对应pool的PlayerID, 保存�?mo.Rosters.
其中 profileRosters 好像没用�?br />

金庆 2019-01-31 10:21 发表评论

试用 go mod

金庆 — Wed, 26 Dec 2018 02:07:00 GMT

试用 go mod

(金庆的专�?2018.12)

Go 1.11 支持 module.

代码不需要在 GOPATH/src 目录下�?br />
先初始化模块�Q�生�?`go.mod`

E:\temp
λ mkdir -p testmod\hello

E:\temp
λ cd testmod\hello\

E:\temp\testmod\hello
λ go mod init github.com/jinq0123/hello
go: creating new go.mod: module github.com/jinq0123/hello

创徏 `hello.go`

package main

import (
   "fmt"
   "rsc.io/quote"
)

func main() {
   fmt.Println(quote.Hello())
}

构徏时报 `golang.org/x/text` �q�不上：(x��)

E:\temp\testmod\hello
λ go build
go: golang.org/x/text@v0.0.0-20170915032832-14c0d48ead0c: unrecog
nized import path "golang.org/x/text" (https fetch: Get https://g
olang.org/x/text?go-get=1: dial tcp 216.239.37.1:443: connectex:
A socket operation was attempted to an unreachable network.)
go: error loading module requirements

`go.mod` ��d��

replace golang.org/x/text => github.com/golang/text v0.3.0

然后构徏��成功了�Q?br />
E:\temp\testmod\hello
λ go build
go: finding github.com/golang/text v0.3.0
go: downloading rsc.io/sampler v1.3.0
go: downloading github.com/golang/text v0.3.0

E:\temp\testmod\hello
λ hello.exe
Hello, world.

如果不加版本��P��则会(x��)报错�Q?br />
go.mod:9: replacement module without version must be directory path (rooted or starting with ./ or ../)

Go 1.11.1 replace �q�有问题�Q�仍�?x��)试图连接原地址。目前版�?1.11.4 可以用�?br />
参考：(x��)
https://github.com/golang/go/wiki/Modules

金庆 2018-12-26 10:07 发表评论

金庆 — Mon, 29 Oct 2018 10:47:00 GMT

gotest 是有�~�存�?br />
(金庆的专�?2018.10)

�?gotest �q�行一个测试，往 mongodb 中插入一条，发现有时灵，有时不灵�?br />
因�ؓ(f��)错误地怀�?mgo 用错了，耗费了不��时间�?br />最�l�发现是因�ؓ(f��) gotest 是有�~�存的，输出的是上次�q�行的结果，但是�q�没有实际运行代码�?br />
�q�行有效是因��Z��码刚改过�Q�测试时�?x��)实际运行�?br />
最�l�也是无意间发现的。给 mgo 开启了调试日志�Q�然后比�?�ơ运行，发现输出是一��L(f��ng)��Q?br />只有一行不同：(x��)

ok mail-server/server 0.519s
ok mail-server/server (cached)

明确昄��了第2�ơ是�~�存。前面运行了几十�ơ都忽略�?cached �q�个输出�?br />
��Z��止�~�存�Q�可加上 -count=1 参数�Q?br />go test -count=1

金庆 2018-10-29 18:47 发表评论

K8s获取NodePort

金庆 — Tue, 04 Sep 2018 07:17:00 GMT

K8s获取NodePort

(金庆的专�?2018.9)

��服务用NodePort暴露到外�|�，为避免端口冲�H�，不指定NodePort,
而是让k8s自动选择一个端口�?br />
$ cat get_node_port.yaml
kind: Service
apiVersion: v1
metadata:
name: jq-service
spec:
type: NodePort
selector:
app: MyApp
ports:
- protocol: TCP
port: 80

$ kubectl apply -f get_node_port.yaml
service "jq-service" configured
$ kubectl describe svc/jq-service
Name:                     jq-service
Namespace:                default
Labels:
Annotations:              kubectl...
Selector:                 app=MyApp
Type:                     NodePort
IP:                       10.104.228.187
Port:                     80/TCP
TargetPort:               80/TCP
NodePort:                 32115/TCP
Endpoints:
Session Affinity:         None
External Traffic Policy: Cluster
Events:

可以看到k8s分配了NodePort 32115�?br />
然后需要获取这个动态的NodePort�Q�以通知客户端连接该端口�?br />
package main

import (
    "context"
    "fmt"
    "log"
    "io/ioutil"

    "github.com/ghodss/yaml"
    "github.com/ericchiang/k8s"
    corev1 "github.com/ericchiang/k8s/apis/core/v1"
)

func main() {
    data, err := ioutil.ReadFile("config")
    if err != nil {
        panic(err)
    }

    // Unmarshal YAML into a Kubernetes config object.
    var config k8s.Config
    if err := yaml.Unmarshal(data, &config); err != nil {
        panic(err)
    }

    client, err := k8s.NewClient(&config)
    // client, err := k8s.NewInClusterClient()
    if err != nil {
        log.Fatal(err)
    }

    var svc corev1.Service
    if err := client.Get(context.Background(), "default", "jq-service", &svc); err != nil {
        log.Fatal(err)
    }
    fmt.Printf("%d\n", *svc.Spec.Ports[0].NodePort)
}

�q�行旉��要复制config: `cp ~/.kube/config .`

金庆 2018-09-04 15:17 发表评论

grpc中的dns负蝲均衡

金庆 — Wed, 29 Aug 2018 02:59:00 GMT

grpc中的dns负蝲均衡

(金庆的专�?2018.8)

grpc-go 中如下连接服务器�Q�请求将在多个IP之间轮�{�?br />
    conn, err := grpc.Dial(
       "dns:///rng-headless:8081",
       grpc.WithBalancerName(roundrobin.Name),
       grpc.WithInsecure())

标准的目标名应该是这��L(f��ng)��Q�`"dns://authority/endpoint_name"`,
此处 authority 为空�Q�详见：(x��)https://github.com/grpc/grpc/blob/master/doc/naming.md

服务器开3个实例，所有请求在3个实例上轮�{�Q?br />
[jinqing@host-10-2-3-4 RoundRobin]$ kubectl run -it --rm jinqing-roundrobin --image=jinq0123/roundrobin:4
If you don't see a command prompt, try pressing enter.
2018/08/28 10:18:01 request 7754383576636566559
2018/08/28 10:18:02 request 2543876599219675746
2018/08/28 10:18:03 request 927204261937181213
2018/08/28 10:18:04 request 7754383576636566559
2018/08/28 10:18:05 request 2543876599219675746
2018/08/28 10:18:06 request 927204261937181213
...

服务器返回一个随机数�Q�不同实例的随机��C��同。代码是�?br />https://github.com/kcollasarundell/balancing-on-k8s 修改的�?br />
...
const (
        port = ":8081"
)

type server struct{}

var r int64

func init(){
    rand.Seed(time.Now().UnixNano())
    r = rand.Int63()
}

func (s *server) Rng(context.Context, *rng.Source) (*rng.RN, error) {
        return &rng.RN{RN: r}, nil
}

func main() {
        lis, err := net.Listen("tcp", port)
        if err != nil {
                log.Fatalf("failed to listen: %v", err)
        }
        s := grpc.NewServer()
        rng.RegisterRngServer(s, &server{})
        // Register reflection service on gRPC server.
        reflection.Register(s)

        if err := s.Serve(lis); err != nil {
                log.Fatalf("failed to serve: %v", err)
        }
}

先编译，打包成镜像，然后�?`balancing-on-k8s\backend\kube.yaml` �q�行�Q?br />kubectl apply -f kube.yaml

`backend\kube.yaml` 创徏了一�?ClusterIP 服务和一�?Headless 服务�Q�部�|�了 3 个服务器实例�?br />[jinqing@host-10-2-3-4 RoundRobin]$ kubectl get svc
NAME           TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)             AGE
kubernetes     ClusterIP   10.96.0.1               443/TCP             93d
rng-cluster    ClusterIP   10.111.30.205           8081/TCP            4h
rng-headless   ClusterIP   None                    8081/TCP,8080/TCP   4h

客户端是一个简单的grpc, 定时发送请求，打印�q�回的随机数�?br />`balancing-on-k8s\clientSideBalancer\RoundRobin\main.go`中的地址需要添加端口，
不然grpc�?x��)去�q�接 443 端口而失败�?br />
扩容后，��到大概3分钟后才看到负蝲转移。羃容后�?x��)立即生效�?br />kubectl scale --replicas=5 deployment/rng

如果�?ClusterIP 服务, 则服务名对应一个ClusterIP;
如果�?Headless 服务�Q�则服务名对应各个Pod的IP:

/ # nslookup rng-headless
Server:         10.96.0.10
Address:        10.96.0.10#53

Name:   rng-headless.default.svc.cluster.local
Address: 10.244.3.27
Name:   rng-headless.default.svc.cluster.local
Address: 10.244.0.108
Name:   rng-headless.default.svc.cluster.local
Address: 10.244.2.66

/ # nslookup rng-cluster
Server:         10.96.0.10
Address:        10.96.0.10#53

Name:   rng-cluster.default.svc.cluster.local
Address: 10.111.30.205

/ #

如果去除 "dns:///", 仅仅是域名加端口�Q?br />
conn, err := grpc.Dial(
       "rng-headless:8081",
       grpc.WithBalancerName(roundrobin.Name),
       ...

则只�?x��)请求同一个实例。只有当该实例pod被删除后才会(x��)切换到另一个实例�?br />使用�~�容时发��C��(x��)优先删除没有客户端连接的实例�?br />�?个客��L(f��ng)��q�接��C��同服务器实例�Q�然后羃容�ؓ(f��)1实例�Q�就可以看到��h��切换�?br />
如果客户端和服务器数量很大，�q�个dns负蝲均衡��׃��合适了�Q�因为客��L(f��ng)��?x��)连接每个服务器实例�?br />
参考：(x��)
Exploring Kubernetes Service Discovery and loadbalancing ( https://kca.id.au/post/k8s_service/ )

金庆 2018-08-29 10:59 发表评论

金庆 — Sun, 22 Jul 2018 03:14:00 GMT

k8s集群外go客户端示�?br />
(金庆的专�?2018.7)

集群内客��L(f��ng)��需要打包成docker镜像�Q�上传镜像，然后�?kubectl run �q�行�Q?br />�q�要讄��用户角色�Q�太�ȝ��Q�还是用集群外客��L(f��ng)��试比较方便�?br />
客户端库使用 ericchiang/k8s, 比官方的 client-go 要简单许多�?br />
集群内客��L(f��ng)��使用`k8s.NewInClusterClient()`创徏�Q?br />集群外客��L(f��ng)��使用 `NewClient(config *Config)`, 需要输入配�|�，
配置��是�?~/.kube/config ��d��的�?br />参�?https://github.com/ericchiang/k8s/issues/79

代码如下�Q?br />
package main

import (
    "context"
    "fmt"
    "log"
    "io/ioutil"

    "github.com/ghodss/yaml"
    "github.com/ericchiang/k8s"
    corev1 "github.com/ericchiang/k8s/apis/core/v1"
)

func main() {
    data, err := ioutil.ReadFile("config")
    if err != nil {
        panic(err)
    }

    // Unmarshal YAML into a Kubernetes config object.
    var config k8s.Config
    if err := yaml.Unmarshal(data, &config); err != nil {
        panic(err)
    }

    client, err := k8s.NewClient(&config)
    // client, err := k8s.NewInClusterClient()
    if err != nil {
        log.Fatal(err)
    }

    var nodes corev1.NodeList
    if err := client.List(context.Background(), "", &nodes); err != nil {
        log.Fatal(err)
    }
    for _, node := range nodes.Items {
        fmt.Printf("name=%q schedulable=%t\n", *node.Metadata.Name, !*node.Spec.Unschedulable)
    }
}

yaml 库用�?ghodss/yaml�Q�不能用 go-yaml, 不然报错
`yaml: unmarshal errors`
见：(x��)https://github.com/ericchiang/k8s/issues/81

复制 .kube/config 到运行目录，�q�行列出所有节点：(x��)

[jinqing@host-10-1-2-19 out-cluster]$ cp ~/.kube/config .
[jinqing@host-10-1-2-19 out-cluster]$ ./out-cluster
name="host-10-1-2-20" schedulable=true
name="host-10-1-2-21" schedulable=true
name="host-10-1-2-22" schedulable=true
name="host-10-1-2-19" schedulable=true

name="host-10-1-2-18" schedulable=true

金庆 2018-07-22 11:14 发表评论

金庆 — Mon, 25 Jun 2018 09:16:00 GMT

用目录结构表�C�go包依赖关�p?br />
(金庆的专�?2018.6)

摘自�Q?br />https://www.ardanlabs.com/blog/2017/02/package-oriented-design.html

If a package wants to import another package at the same level:

* Question the current design choices of these packages.
* If reasonable, move the package inside the source tree for the package that wants to import it.
* Use the source tree to show the dependency relationships.

金庆 2018-06-25 17:16 发表评论

金庆 — Mon, 25 Jun 2018 08:51:00 GMT

Golang的包�?br />
(金庆的专�?2018.6)

摘自�Q?br />
https://talks.golang.org/2014/organizeio.slide#1

The name of a package

Keep package names short and meaningful.
Don't use underscores, they make package names long.

    io/ioutil not io/util
    suffixarray not suffix_array

Don't overgeneralize. A util package could be anything.

The name of a package is part of its type and function names.
On its own, type Buffer is ambiguous. But users see:

    buf := new(bytes.Buffer)

Choose package names carefully.

Choose good names for users.

金庆 2018-06-25 16:51 发表评论

grpc-go与actor模式

金庆 — Tue, 12 Jun 2018 03:15:00 GMT

grpc-go与actor模式

(金庆的专�?2018.6)

grpc-go服务器的每个��h��都在一个独立的协程中执行�?br />�|�游服务器中�Q�一般请求会(x��)调用游戏戉K��的方法，而房间是一个独立的协程�?br />可以��房间实��Cؓ(f��)actor�Q�grpc��h��通过Call()或Post()�Ҏ(gu��)��来执行�?br />其中Call()�?x��)等待返回，而Post()�?x��)异步执行无�q�回倹{�?br />
type Room struct {
   // actC 是其他协�E�向Room协程发送动作的Channel�Q�协�E�中��依�ơ执行动作�?/span>
   // Action 动作, 是无参数无返回值的函数.
   actC chan func()

   ...
}

// Run �q�行戉K��协程.
func (r *Room) Run() {
   ticker := time.NewTicker(20 * time.Millisecond)
   defer ticker.Stop()

   for r.running {
       select {
       case act := <-r.actC:
           act()
       case <-ticker.C:
           r.tick()
       }
   }
}

// Call calls a function f and returns the result.
// f runs in the Room's goroutine.
func (r *Room) Call(f func() interface{}) interface{} {
   // �l�果从ch�q�回
   ch := make(chan interface{}, 1)
   r.actC <- func() {
       ch <- f()
   }
   // �{�待直到�q�回�l�果
   return <-ch
}

// Post ��一个动作投递到内部协程中执�?
func (r *Room) Post(f func()) {
   r.actC <- f
}

grpc服务�Ҏ(gu��)��如：(x��)

func (m *RoomService) Test(ctx context.Context, req *pb.TestReq) (*pb.TestResp, error) {
   conn := conn_mgr.GetConn(ctx)
   if conn == nil {
       return nil, fmt.Errorf("can not find connection")
   }

   room := conn.GetRoom()
   resp := room.Call(func() interface{} {
       return room.Test(req)
   })

   return resp.(*pb.TestResp), nil
}

金庆 2018-06-12 11:15 发表评论

精品乱码久久久久久夜夜嗨,久久久久久亚洲Av无码精品专口,久久99国产精品二区不卡

net.LookupSRV()查询k8s无头服务

rpc应答太快造成��h�����时

数组make参数错误

go不要导出channel

go代码覆盖���试

# go代码覆盖���试

# 考察go一致性hash�?/h1>(金庆的专�?2020.2)github �?consistent, 按星��C���ơ察�?br />

## stathat/consistent

## lafikl/consistent

## serialx/hashring

## buraksezer/consistent

## 其他

### dgryski/go-jump

### lithammer/go-jump-consistent-hash

���go函数指针转�ؓ(f��)接口

# ���go函数指针转�ؓ(f��)接口

golang chan 关闭时的原则

open-match匚w�����程

# open-match匚w�����程

## test/cmd/frontendclient

## examples/backendclient

## cmd/frontendapi

## cmd/backendapi

## cmd/mmforc

## mmf

### �l�果

### �l�节

试用 go mod

K8s获取NodePort

grpc中的dns负蝲均衡

grpc-go与actor模式

rpc应答太快造成��h��时

go代码覆盖��试

# go代码覆盖��试

# 考察go一致性hash�?/h1>
(金庆的专�?2020.2)

github �?consistent, 按星��C��ơ察�?br />

��go函数指针转�ؓ(f��)接口

# ��go函数指针转�ؓ(f��)接口

open-match匚w��程

# open-match匚w��程