数据湖（十六）：Structured Streaming实时写入Iceberg

[root@node1 bin]# ./kafka-topics.sh  --zookeeper node3:2181,node4:2181,node5:2181  --create  --topic kafka-iceberg-topic  --partitions 3 --replication-factor 3

/**
  * 向Kafka中写入数据
  */
object WriteDataToKafka{
  def main(args: Array[String]): Unit = {
    val props = new Properties()
    props.put("bootstrap.servers", "node1:9092,node2:9092,node3:9092")
    props.put("key.serializer", "org.apache.kafka.common.serialization.StringSerializer")
    props.put("value.serializer", "org.apache.kafka.common.serialization.StringSerializer")

    val producer = new KafkaProducer[String,String](props)
    var counter = 0
    var keyFlag = 0
    while(true){
      counter +=1
      keyFlag +=1
      val content: String = userlogs()
      producer.send(new ProducerRecord[String, String]("kafka-iceberg-topic", content))
      //producer.send(new ProducerRecord[String, String]("kafka-iceberg-topic", s"key-$keyFlag", content))
      if(0 == counter%100){
        counter = 0
        Thread.sleep(5000)
      }
    }
    producer.close()
  }

  def userlogs()={
    val userLogBuffer = new StringBuffer("")
    val timestamp = new Date().getTime();
    var userID = 0L
    var pageID = 0L

    //随机生成的用户ID
    userID = Random.nextInt(2000)

    //随机生成的页面ID
    pageID =  Random.nextInt(2000);

    //随机生成Channel
    val channelNames = Array[String]("Spark","Scala","Kafka","Flink","Hadoop","Storm","Hive","Impala","HBase","ML")
    val channel = channelNames(Random.nextInt(10))

    val actionNames = Array[String]("View", "Register")
    //随机生成action行为
    val action = actionNames(Random.nextInt(2))

    val dateToday = new SimpleDateFormat("yyyy-MM-dd").format(new Date())
    userLogBuffer.append(dateToday)
      .append("\t")
      .append(timestamp)
      .append("\t")
      .append(userID)
      .append("\t")
      .append(pageID)
      .append("\t")
      .append(channel)
      .append("\t")
      .append(action)
    System.out.println(userLogBuffer.toString())
    userLogBuffer.toString()
  }
}

object StructuredStreamingSinkIceberg{
  def main(args: Array[String]): Unit = {
    //1.准备对象
    val spark: SparkSession = SparkSession.builder().master("local").appName("StructuredSinkIceberg")
      //指定hadoop catalog，catalog名称为hadoop_prod
      .config("spark.sql.catalog.hadoop_prod", "org.apache.iceberg.spark.SparkCatalog")
      .config("spark.sql.catalog.hadoop_prod.type", "hadoop")
      .config("spark.sql.catalog.hadoop_prod.warehouse", "hdfs://mycluster/structuredstreaming")
      .getOrCreate()
//    spark.sparkContext.setLogLevel("Error")

    //2.创建Iceberg 表
    spark.sql(
      """
        |create table if not exists hadoop_prod.iceberg_db.iceberg_table (
        | current_day string,
        | user_id string,
        | page_id string,
        | channel string,
        | action string
        |) using iceberg
      """.stripMargin)

    val checkpointPath = "hdfs://mycluster/iceberg_table_checkpoint"
    val bootstrapServers = "node1:9092,node2:9092,node3:9092"
    //多个topic 逗号分开
    val topic = "kafka-iceberg-topic"

    //3.读取Kafka读取数据
    val df = spark.readStream
      .format("kafka")
      .option("kafka.bootstrap.servers", bootstrapServers)
      .option("auto.offset.reset", "latest")
      .option("group.id", "iceberg-kafka")
      .option("subscribe", topic)
      .load()

    import spark.implicits._
    import org.apache.spark.sql.functions._

    val resDF = df.selectExpr("CAST(key AS STRING)", "CAST(value AS STRING)")
      .as[(String, String)].toDF("id", "data")

    val transDF: DataFrame = resDF.withColumn("current_day", split(col("data"), "\t")(0))
      .withColumn("ts", split(col("data"), "\t")(1))
      .withColumn("user_id", split(col("data"), "\t")(2))
      .withColumn("page_id", split(col("data"), "\t")(3))
      .withColumn("channel", split(col("data"), "\t")(4))
      .withColumn("action", split(col("data"), "\t")(5))
      .select("current_day", "user_id", "page_id", "channel", "action")

    //结果打印到控制台,Default trigger (runs micro-batch as soon as it can)
//    val query: StreamingQuery = transDF.writeStream
//      .outputMode("append")
//      .format("console")
//      .start()

    //4.流式写入Iceberg表
    val query = transDF.writeStream
      .format("iceberg")
      .outputMode("append")
      //每分钟触发一次Trigger.ProcessingTime(1, TimeUnit.MINUTES)
      //每10s 触发一次 Trigger.ProcessingTime(1, TimeUnit.MINUTES)
      .trigger(Trigger.ProcessingTime(10, TimeUnit.SECONDS))
      .option("path", "hadoop_prod.iceberg_db.iceberg_table")
      .option("fanout-enabled", "true")
      .option("checkpointLocation", checkpointPath)
      .start()

    query.awaitTermination()

  }
}

//1.准备对象
val spark: SparkSession = SparkSession.builder().master("local").appName("StructuredSinkIceberg")
  //指定hadoop catalog，catalog名称为hadoop_prod
  .config("spark.sql.catalog.hadoop_prod", "org.apache.iceberg.spark.SparkCatalog")
  .config("spark.sql.catalog.hadoop_prod.type", "hadoop")
  .config("spark.sql.catalog.hadoop_prod.warehouse", "hdfs://mycluster/structuredstreaming")
  .getOrCreate()

//2.读取Iceberg 表中的数据结果
spark.sql(
  """
    |select * from hadoop_prod.iceberg_db.iceberg_table
  """.stripMargin).show()