GoLang RabbitMQ实现六种工作模式示例
Onemorelight95 人气:0六种工作模式介绍
1.简单(Simple)模式
P:生产者,也就是要发送消息的程序。
C:消费者:消息的接收者,会一直等待消息到来。
queue:消息队列,图中红色部分。类似一个邮箱,可以缓存消息;生产者向其中投递消息,消费者从其中取出消息。
简单模式就是单发单收,消息的消费者监听消息队列,如果队列中有消息,就消费掉,消息被拿走后,自动从队列中删除。
2.工作队列(Work Queue)模式
这种模式就是多个消费者消费同一个队列中的消息,既然消费者多了那么就出现了消息分配的问题,所以对应着两种分配策略:
- 公平分发:每个消费者接收消息的概率是相等的,消息队列会循环依次给每个消费者发送消息,这种是默认的策略。
- 公平派遣:保证消费者在消费完某个消息,并发送确认信息后,消息队列才会向它推送新的消息,在此之间若是有新的消息,将会被推送到其它消费者,若所有的消费者都在消费消息,那么就会等待。
3.发布/订阅(Pub/Sub)模式
在这种模型中,多了一个 Exchange 角色,而且过程略有变化:
P:生产者,也就是要发送消息的程序,但是不再发送到队列中,而是发给X (交换机)。
C:消费者,消息的接收者,会一直等待消息到来。
Queue:消息队列,接收消息、缓存消息。
Exchange:交换机(X) ,一方面,接收生产者发送的消息。另一方面,如何处理消息,递交给某个特别队列、递交给所有队列、或是将消息丢弃。到底如何操作,取决于Exchange的类型。 Exchange有以下4种类型:
- Fanout:广播,将消息交给所有绑定到交换机的队列。
- Direct:全值匹配,把消息交给符合指定
routing key
的队列。 - Topic:通配符,与Direct类型类似,但Direct类型要求
routing key
完全相等,而Topic类型是对routing key
进行模糊匹配,比Direct灵活。 - Headers:根据Message的一些头部信息来分发过滤Message,用的比较少。
注意:Exchange负责转发消息,不具备存储消息的能力,因此如果没有任何队列与Exchange绑定,或者没有符合路由规则的队列,那么消息会丢失。
4.路由(Routing)模式
路由模式其实就是上述发布/订阅模式的交换机转发类型变成了Direct类型。在这种模式下:
Exchange 不再把消息交给每一个绑定的队列,而是根据消息的routing key
进行判断,只有队列的
routing key
与消息的routing key
完全一致,才会接收到消息。
P:生产者,向 Exchange 发送消息,发送消息时,会指定一个routing key
。
X:Exchange(交换机),接收生产者的消息,然后把消息递交给与routing key
完全匹配的队列。
C1:消费者,其所在队列指定了需要routing key
为error的消息。
C2:消费者,其所在队列指定了需要routing key
为 info、error、warning 的消息。
5.通配符(Tpoic)模式
路由模式其实就是上述发布/订阅模式的交换机转发类型变成了Topic类型。在这种模式下:
队列的routing key
与消息的routing key
符合匹配规则,就可以接收到消息,有两种规则:
*
:可以(只能)匹配一个单词。
#
:可以匹配多个单词(或者零个)。
所以图中,routing key
为a.orange.b
的消息就会被转发到Q1,而routing key
为Lazy.a.b.c
的消息就会被发送到Q2。
Go语言的实现
安装操作库
安装API库
Go可以使用streadway/amqp
库来操作rabbit,使用以下命令来安装:
go get github.com/streadway/amqp
封装rabbitmq
接下来我们对streadway/amqp
库的内容进行一个二次封装,封装为一个rabbitmq.go
文件:
package rabbitmq import ( "encoding/json" "github.com/streadway/amqp" "log" ) // RabbitMQ RabbitMQ结构 type RabbitMQ struct { channel *amqp.Channel Name string exchange string } // Connect 连接服务器 func Connect(s string) *RabbitMQ { //连接rabbitmq conn, e := amqp.Dial(s) failOnError(e, "连接Rabbitmq服务器失败!") ch, e := conn.Channel() failOnError(e, "无法打开频道!") mq := new(RabbitMQ) mq.channel = ch return mq } // New 初始化消息队列 //第一个参数:rabbitmq服务器的链接,第二个参数:队列名字 func New(s string, name string) *RabbitMQ { //连接rabbitmq conn, e := amqp.Dial(s) failOnError(e, "连接Rabbitmq服务器失败!") ch, e := conn.Channel() failOnError(e, "无法打开频道!") q, e := ch.QueueDeclare( name, //队列名 false, //是否开启持久化 true, //不使用时删除 false, //排他 false, //不等待 nil, //参数 ) failOnError(e, "初始化消息队列失败!") mq := new(RabbitMQ) mq.channel = ch mq.Name = q.Name return mq } // QueueDeclare 声明queue func (q *RabbitMQ) QueueDeclare(queue string) { _, e := q.channel.QueueDeclare(queue, false, true, false, false, nil) failOnError(e, "声明queue失败!") } // QueueDelete 删除queue func (q *RabbitMQ) QueueDelete(queue string) { _, e := q.channel.QueueDelete(queue, false, true, false) failOnError(e, "删除queue失败!") } // Qos 配置queue参数 func (q *RabbitMQ) Qos() { e := q.channel.Qos(1, 0, false) failOnError(e, "无法设置QoS") } // NewExchange 初始化交换机 //第一个参数:rabbitmq服务器的链接,第二个参数:交换机名字,第三个参数:交换机类型 func NewExchange(s string, name string, typename string) { //连接rabbitmq conn, e := amqp.Dial(s) failOnError(e, "连接Rabbitmq服务器失败!") ch, e := conn.Channel() failOnError(e, "无法打开频道!") e = ch.ExchangeDeclare( name, // name typename, // type true, // durable false, // auto-deleted false, // internal false, // no-wait nil, // arguments ) failOnError(e, "初始化交换机失败!") } // ExchangeDelete 删除交换机 func (q *RabbitMQ) ExchangeDelete(exchange string) { e := q.channel.ExchangeDelete(exchange, false, true) failOnError(e, "删除交换机失败!") } // Bind 绑定消息队列到exchange func (q *RabbitMQ) Bind(exchange string, key string) { e := q.channel.QueueBind( q.Name, key, exchange, false, nil, ) failOnError(e, "绑定队列失败!") q.exchange = exchange } // Send 向消息队列发送消息 //Send方法可以往某个消息队列发送消息 func (q *RabbitMQ) Send(queue string, body interface{}) { str, e := json.Marshal(body) failOnError(e, "消息序列化失败!") e = q.channel.Publish( "", //交换 queue, //路由键 false, //必填 false, //立即 amqp.Publishing{ ReplyTo: q.Name, Body: []byte(str), }) msg := "向队列:" + q.Name + "发送消息失败!" failOnError(e, msg) } // Publish 向exchange发送消息 //Publish方法可以往某个exchange发送消息 func (q *RabbitMQ) Publish(exchange string, body interface{}, key string) { str, e := json.Marshal(body) failOnError(e, "消息序列化失败!") e = q.channel.Publish( exchange, key, false, false, amqp.Publishing{ReplyTo: q.Name, Body: []byte(str)}, ) failOnError(e, "向交换机发送消息失败!") } // Consume 接收某个消息队列的消息 func (q *RabbitMQ) Consume() <-chan amqp.Delivery { c, e := q.channel.Consume( q.Name, //指定从哪个队列中接收消息 "", true, false, false, false, nil, ) failOnError(e, "接收消息失败!") return c } // Close 关闭队列连接 func (q *RabbitMQ) Close() { q.channel.Close() } //错误处理函数 func failOnError(err error, msg string) { if err != nil { log.Fatalf("%s: %s", msg, err) } }
简单(Simple)模式
生产者
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 producer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "simple") i := 0 for { // 每隔2s发送一次消息 time.Sleep(time.Second * 2) producer.Send("simple", " simple message: "+strconv.Itoa(i)) i = i + 1 } }
消费者
func main() { consumer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "simple") //接收消息时,指定 messages := consumer.Consume() go func() { for ch := range messages { log.Printf("Received a message: %s", ch.Body) // 消费消息要用3s time.Sleep(time.Second * 3) } }() select {} }
运行结果:
2022/11/05 18:54:47 Received a message: " simple message: 0"
2022/11/05 18:54:52 Received a message: " simple message: 1"
2022/11/05 18:54:57 Received a message: " simple message: 2"
工作队列(Work Queue)模式
公平分发模式:
公平分发模式采用的是轮询机制,它会将数个任务按顺序平均分发给消费者。
生产者
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 producer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "worker") i := 0 for { // 每隔2s发送一次消息 time.Sleep(time.Second * 2) producer.Send("worker", " worker message: "+strconv.Itoa(i)) i = i + 1 } }
消费者1
func main() { consumer1 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "worker") //接收消息 messages := consumer1.Consume() go func() { for ch := range messages { log.Printf("Received a message: %s", ch.Body) // 消费消息要用3s time.Sleep(time.Second * 3) } }() select {} }
消费者2
func main() { consumer2 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "worker") //接收消息 messages := consumer2.Consume() go func() { for ch := range messages { log.Printf("Received a message: %s", ch.Body) // 消费消息要用3s time.Sleep(time.Second * 3) } }() select {} }
运行结果:
# 消费者1
2022/11/05 19:45:03 Received a message: " worker message: 0"
2022/11/05 19:45:07 Received a message: " worker message: 2"
2022/11/05 19:45:11 Received a message: " worker message: 4"# 消费者2
2022/11/05 19:45:05 Received a message: " worker message: 1"
2022/11/05 19:45:09 Received a message: " worker message: 3"
2022/11/05 19:45:13 Received a message: " worker message: 5"
可以发现,公平模式下,偶数消息都被发送给了消费者1,而奇数消息都被发送给了消费者2。
公平派遣模式:
有时候,如果消息之间的复杂度不同,那么不同消费者消费消息所用的时间会不同。这个时候如果使用公平派发模式,可能会造成某一个消费者需要消费的消息积压过多。可以采用公平派遣模式:
公平派遣模式下发送端与公平分发相同,消费者端只需要加一段配置代码,我们可以将预取计数设置为1。这告诉RabbitMQ一次不要给消费者一个以上的消息。换句话说,在处理并确认上一条消息之前,不要将新消息发送给消费者。而是将其分派给不忙的下一个消费者。
关于消息的确认:
为了确保消息永不丢失,RabbitMQ支持 消息确认。消费者发送回一个确认(acknowledgement),以告知RabbitMQ已经接收,处理了特定的消息,并且RabbitMQ可以自由删除它。
我们之前的代码中,RabbitMQ一旦向消费者传递了一条消息,便立即将其标记为删除(调用Consumer的第三个参数是autoAck,表示是否自动回复)。在这种情况下,如果你终止一个消费者那么你就可能会丢失这个任务,我们还将丢失所有已经交付给这个消费者的尚未消费的消息。如果一个消费者意外宕机了,那么我们希望将任务交付给其他消费者来消费者。
所以一旦向消费者传递了一条消息,就不能马上将其标记为删除,而是要手动确认。我们需要在创建消费者的时候将autoAck
参数标记为false:
// Consume 接收某个消息队列的消息 func (q *RabbitMQ) Consume() <-chan amqp.Delivery { c, e := q.channel.Consume( q.Name, //指定从哪个队列中接收消息 "", false, // 不自动确认消息 false, false, false, nil, ) failOnError(e, "接收消息失败!") return c }
然后每消费完一条消息需要调用Ack(false)
函数手动回复。
生产者
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 producer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "worker") i := 0 for { // 每隔2s发送一次消息 time.Sleep(time.Second * 2) producer.Send("worker", " worker message: "+strconv.Itoa(i)) i = i + 1 } }
消费端限流:
实现公平派遣模式我们需要设置消费者端一次只能消费一条消息,之前我们已经进行了封装,直接在消费者端调用即可:
// Qos 配置queue参数 func (q *RabbitMQ) Qos() { e := q.channel.Qos(1, 0, false) failOnError(e, "无法设置QoS") }
消费者1
func main() { consumer1 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "worker") // 指定一次只消费一条消息,直到消费完才重新接收 consumer1.Qos() //接收消息 messages := consumer1.Consume() go func() { for ch := range messages { log.Printf("Received a message: %s", ch.Body) // 消费消息要用10s time.Sleep(time.Second * 10) // 手动回复 ch.Ack(false) } }() select {} }
消费者2
func main() { consumer2 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "worker") // 指定一次只消费一条消息,直到消费完才重新接收 consumer2.Qos() //接收消息 messages := consumer2.Consume() go func() { for ch := range messages { log.Printf("Received a message: %s", ch.Body) // 消费消息要用2s time.Sleep(time.Second * 2) // 手动回复 ch.Ack(false) } }() select {} }
运行结果:
# 消费者1
2022/11/05 20:31:26 Received a message: " worker message: 0"
2022/11/05 20:31:36 Received a message: " worker message: 5"# 消费者2
2022/11/05 20:31:28 Received a message: " worker message: 1"
2022/11/05 20:31:30 Received a message: " worker message: 2"
2022/11/05 20:31:32 Received a message: " worker message: 3"
2022/11/05 20:31:34 Received a message: " worker message: 4"
2022/11/05 20:31:38 Received a message: " worker message: 6"
2022/11/05 20:31:40 Received a message: " worker message: 7"
发布/订阅(Pub/Sub)模式
生产者
func main() { producer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue") rabbitmq.NewExchange("amqp://guest:guest@35.76.111.125:5672/", "exchange1", "fanout") i := 0 for { time.Sleep(time.Second) // fanout模式下不用routing key producer.Publish("exchange1", "pubsub message: "+strconv.Itoa(i), "") i = i + 1 } }
消费者1
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 consumer1 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue1") // 队列绑定到exchange consumer1.Bind("exchange1", "") //接收消息 msgs := consumer1.Consume() go func() { for d := range msgs { log.Printf("Consumer1 received a message: %s", d.Body) d.Ack(false) } }() select {} }
消费者2
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 consumer2 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue2") // 队列绑定到exchange consumer2.Bind("exchange1", "") //接收消息 msgs := consumer2.Consume() go func() { for d := range msgs { log.Printf("Consumer2 received a message: %s", d.Body) d.Ack(false) } }() select {} }
运行结果:
# 消费者1
2022/11/05 22:32:19 Consumer1 received a message: "pubsub message: 0"
2022/11/05 22:32:20 Consumer1 received a message: "pubsub message: 1"
2022/11/05 22:32:21 Consumer1 received a message: "pubsub message: 2"
2022/11/05 22:32:22 Consumer1 received a message: "pubsub message: 3"
2022/11/05 22:32:23 Consumer1 received a message: "pubsub message: 4"
2022/11/05 22:32:24 Consumer1 received a message: "pubsub message: 5"# 消费者2
2022/11/05 22:32:19 Consumer2 received a message: "pubsub message: 0"
2022/11/05 22:32:20 Consumer2 received a message: "pubsub message: 1"
2022/11/05 22:32:21 Consumer2 received a message: "pubsub message: 2"
2022/11/05 22:32:22 Consumer2 received a message: "pubsub message: 3"
2022/11/05 22:32:23 Consumer2 received a message: "pubsub message: 4"
2022/11/05 22:32:24 Consumer2 received a message: "pubsub message: 5"
路由(Routing)模式
生产者
func main() { producer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue") // 指定为direct类型 rabbitmq.NewExchange("amqp://guest:guest@35.76.111.125:5672/", "exchange", "direct") i := 0 for { time.Sleep(time.Second) // 如果是奇数,就发key1 // 如果是偶数,就发key2 if i%2 != 0 { producer.Publish("exchange", "routing message: "+strconv.Itoa(i), "key1") } else { producer.Publish("exchange", "routing message: "+strconv.Itoa(i), "key2") } i = i + 1 } }
消费者1
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 consumer1 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue1") // 队列绑定到exchange consumer1.Bind("exchange", "key1") //接收消息 msgs := consumer1.Consume() go func() { for d := range msgs { log.Printf("Consumer1 received a message: %s", d.Body) d.Ack(false) } }() select {} }
消费者2
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 consumer2 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue2") // 队列绑定到exchange consumer2.Bind("exchange", "key2") //接收消息 msgs := consumer2.Consume() go func() { for d := range msgs { log.Printf("Consumer2 received a message: %s", d.Body) d.Ack(false) } }() select {} }
运行结果:
# 消费者1
2022/11/05 22:51:10 Consumer1 received a message: "routing message: 1"
2022/11/05 22:51:12 Consumer1 received a message: "routing message: 3"
2022/11/05 22:51:14 Consumer1 received a message: "routing message: 5"
2022/11/05 22:51:16 Consumer1 received a message: "routing message: 7"# 消费者2
2022/11/05 22:51:11 Consumer2 received a message: "routing message: 0"
2022/11/05 22:51:13 Consumer2 received a message: "routing message: 2"
2022/11/05 22:51:15 Consumer2 received a message: "routing message: 4"
2022/11/05 22:51:17 Consumer2 received a message: "routing message: 6"
通配符(Tpoic)模式
生产者
func main() { producer := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue") // 指定为topic类型 rabbitmq.NewExchange("amqp://guest:guest@35.76.111.125:5672/", "exchange2", "topic") var i int for { time.Sleep(time.Second) if i%2 != 0 { producer.Publish("exchange2", "topic message: "+strconv.Itoa(i), "a.test.b.c") } else { producer.Publish("exchange2", "topic message: "+strconv.Itoa(i), "a.test.b") } i++ } }
消费者1
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 consumer1 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue1") // 队列绑定到exchange consumer1.Bind("exchange2", "*.test.*") //接收消息 msgs := consumer1.Consume() go func() { for d := range msgs { log.Printf("Consumer1 received a message: %s", d.Body) d.Ack(false) } }() select {} }
消费者2
func main() { //第一个参数指定rabbitmq服务器的链接,第二个参数指定创建队列的名字 consumer2 := rabbitmq.New("amqp://guest:guest@35.76.111.125:5672/", "queue2") // 队列绑定到exchange consumer2.Bind("exchange2", "#.test.#") //接收消息 msgs := consumer2.Consume() go func() { for d := range msgs { log.Printf("Consumer2 received a message: %s", d.Body) d.Ack(false) } }() select {} }
运行结果:
# 消费者1
2022/11/05 23:09:53 Consumer1 received a message: "topic message: 0"
2022/11/05 23:09:55 Consumer1 received a message: "topic message: 2"
2022/11/05 23:09:57 Consumer1 received a message: "topic message: 4"
2022/11/05 23:09:59 Consumer1 received a message: "topic message: 6"# 消费者2
2022/11/05 23:09:53 Consumer2 received a message: "topic message: 0"
2022/11/05 23:09:54 Consumer2 received a message: "topic message: 1"
2022/11/05 23:09:55 Consumer2 received a message: "topic message: 2"
2022/11/05 23:09:56 Consumer2 received a message: "topic message: 3"
2022/11/05 23:09:57 Consumer2 received a message: "topic message: 4"
2022/11/05 23:09:58 Consumer2 received a message: "topic message: 5"
2022/11/05 23:09:59 Consumer2 received a message: "topic message: 6"
加载全部内容