Flink 消费 Kafka 数据批量写入 MySQL 多个表

最新推荐文章于 2024-07-27 15:05:24 发布

訾零

最新推荐文章于 2024-07-27 15:05:24 发布

阅读量4k

点赞数 3

分类专栏： Flink

本文链接：https://blog.csdn.net/lingeio/article/details/105238515

版权

业务场景：
sdk 数据统一入Kafka 的一个 topic（topic_sdk_log），其中包含多种事件类型数据，如：登录，注册，激活等，需要将 Kafka 中数据根据事件类型分别写入 MySQL 多个表。这里使用 Flink 每5秒写入 MySQL 不同表。

数据示例：

{"key":"login","data":{"_game_version":"","_package_id":"12280200","_core_account":"ysdk_oGFjT0XEjdefIOgd7uApwWUX2ccY","_time":"2020-03-23 15:35:11","_time_server":"2020-03-23 15:35:59","_idfa":"","_imei":"","_mac":null,"_imsi":null,"_device_code":"46e0fd235df0b7c167c0cbe82be02a3c"}}

{"key":"activate","data":{"_game_version":"","_package_id":850030,"_time":"2020-03-23 15:25:56","_time_server":"2020-03-23 15:27:42","_idfa":"","_imei":"866092032174377","_mac":"02:00:00:00:00:00","_imsi":"460001960611910","_device_code":"fc9b4de27db42c94433a02c59cd5e2ff"}}

KafkaToMySQL 主类

import com.alibaba.fastjson.JSON
import com.sm.conf.ConfigurationManager
import com.sm.constants.Constants
import com.sm.sink.MySQLSink
import com.sm.entry.SdkData
import com.sm.util.KafkaUtils
import com.sm.window.SdkWindow
import org.apache.flink.streaming.api.scala._
import org.apache.flink.streaming.api.windowing.time.Time
import org.apache.flink.streaming.api.TimeCharacteristic


/**
  * Flink 读取 Kafka，每分钟聚合一次数据，批量写入 MySQL
  *
  * create by LiuJinHe 2020/3/25
  */
object KafkaToMySQL {
  def main(args: Array[String]): Unit = {
    val env = StreamExecutionEnvironment.getExecutionEnvironment
    // 失败重启
    env.setRestartStrategy(RestartStrategies.fixedDelayRestart(10, 3L))

    //    env.setParallelism(1)

    // checkpoint 周期
    env.enableCheckpointing(10000)
    val config = env.getCheckpointConfig

    // exactly_one 模式
    config.setCheckpointingMode(CheckpointingMode.EXACTLY_ONCE)
    // checkpoint 之间最小间隔
    config.setMinPauseBetweenCheckpoints(1000)
    // checkpoint 超时时间,超时将被丢弃
    config.setCheckpointTimeout(10000)
    // 同一时间只允许进行一次 checkpoint
    config.setMaxConcurrentCheckpoints(1)

    val kafkaProp = ConfigurationManager.load(Constants.KAFKA_PROPERTIES)
    val kafkaConsumer = KafkaUtils.getConsumer(kafkaProp).setStartFromGroupOffsets()

    val dataStream = env.addSource(kafkaConsumer)

    dataStream.map(line => {
      val data = JSON.parseObject(line, classOf[SdkData])
      data
    })
      .filter(line =>
        line.key == "activate" || line.key == "register" || line.key == "start_app" || line.key == "login"
      )
      .keyBy(data => data.key)
      .timeWindowAll(Time.seconds(3))
      .apply(new SdkWindow)
      .addSink(new MySink)

    env.execute("stream