Skip to main content

Kafka 使用

kafka是Java写的, 官方并没有Go SDK, 但是社区有3款还不错的SDK可供选择

  • confluent-kafka-go: 基于c库librdkafka的封装, 文档不错, 但是不支持Go Context
  • sarama: 迄今为止最流行的一个库, 纯Go实现, 但是暴露的API偏低层(Kafka protocol), 使用手感欠佳, 也不支持Go Context
  • kafka-go: 新贵,借鉴了sarama,并且兼容Sarama, 纯Go实现, 代码质量也比之前2个库好, API的封装非常友好, 非常符合Go的编程习惯, 比如Context, Reader, Writer等

这里选择kafka-go

配置组件​

[kafka]
  brokers = ["127.0.0.1:9092"]
  scram_algorithm = "SHA512"
  username = ""
  password = ""
  debug = false

环境​

这里环境采用Docker composej安装:

创建dock compose编排文件: docker-compose.yml

version: '2'
services:
  zookeeper:
    image: wurstmeister/zookeeper
    restart: unless-stopped
    hostname: zookeeper
    ports:
      - "2181:2181"
    container_name: zookeeper

  kafka:
    image: wurstmeister/kafka
    ports:
      - "9092:9092"
    environment:
      KAFKA_ADVERTISED_HOST_NAME: localhost
      KAFKA_ZOOKEEPER_CONNECT: "zookeeper:2181"
      KAFKA_BROKER_ID: 1
      KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR: 1
      KAFKA_CREATE_TOPICS: "stream-in:1:1,stream-out:1:1"
    depends_on:
      - zookeeper
    container_name: kafka

启动kafka服务

docker-compose up -d

基本使用​

package main

import (
	"fmt"

	"github.com/infraboard/mcube/v2/ioc/config/kafka"
)

func main() {
	// 消息生产者
	producer := kafka.Producer("test")
	fmt.Println(producer)

	// 消息消费者
	consumer := kafka.ConsumerGroup("group id", []string{"topic name"})
	fmt.Println(consumer)
}