//flink中关于时间的三个概念 //event time:数据产生的时间 //processing time:处理数据的时间 即操作数据的之间 比如一个flink在scala中的map函数处理数据时 //ingest time:摄取数据时间,在一个streaming程序中 一个时间段收集数据的时间 //而evet time在处理实时数据时是比较有用的,例如在由于网络的繁忙的原因,某些数据未能按时到达,假设迟到了30min, //而我们定义的最大延迟不能超过十分钟,那么一些数据包含了超时的数据那么这些数据是不会在这次操作中处理的而是会 //丢弃掉
//kafka生产者代码
package kafka.partition.test;
import java.util.HashMap;
import java.util.Map;
import org.apache.kafka.clients.producer.KafkaProducer;
import org.apache.kafka.clients.producer.ProducerRecord;
public class PartitionProducer {
public static void main(String[] args) {
Map<String,Object> props = new HashMap<>();
props.put("acks", "1");
props.put("partitioner.class", "org.apache.kafka.clients.producer.internals.DefaultPartitioner");
props.put("key.serializer", "org.apache.kafka.common.serialization.StringSerializer");
props.put("value.serializer", "org.apache.kafka.common.serialization.StringSerializer");
props.put("bootstrap.servers", "bigdata01:9092");
String topic = "event_time";
KafkaProducer<String, String> producer = new KafkaProducer<>(props);
for(int i = 0 ; i <= 20;i++) {
//flink的watermarkassginer里面定义的超时时间是5000毫秒
long mills = System.currentTimeMillis();
if(i%3==0) {
//数据的event time放在字符串的开头 以空格分割
//kafka event_time topic的0分区超时4000毫秒
String line = (mills-4000)+" "+"partition-0--this is a big +" +i;
ProducerRecord< String,String> record = new ProducerRecord<String, String>(topic, new Integer(0), null, i+"", line);
producer.send(record);
}else if(i%3==1) {
//kafka event_time topic的1分区超时5000毫秒
String line = (mills-5000)+" "+"partition-1--this is a big +" +i;
ProducerRecord< String,String> record = new ProducerRecord<String, String>(topic, new Integer(1), null, i+"", line);
producer.send(record);
}else if(i%3==2) {
//kafka event_time topic的2分区超时8000毫秒
String line = (mills-8000)+" "+"partition-2--this is a big +" +i;
ProducerRecord< String,String> record = new ProducerRecord<String, String>(topic, new Integer(2), null, i+"", line);
producer.send(record);
}
}
producer.close();
}
}
//自定义的TimestampsAndWatermarks package flink.streaming import org.apache.flink.streaming.api.functions.AssignerWithPeriodicWatermarks import org.apache.flink.streaming.api.watermark.Watermark class CustomWaterMarks extends AssignerWithPeriodicWatermarks[String]{ //超时时间 val maxOutOrderness = 5000l //flink过一段时间便会调一次该函数获取水印 def getCurrentWatermark():Watermark ={ val sysMilssecons = System.currentTimeMillis() new Watermark(sysMilssecons-maxOutOrderness) } //每条记录多会调用 来获得even time 在生产的数据中 even_time放在字符串的第一个字段 用空格分割 def extractTimestamp(element: String,previousElementTimestamp: Long): Long = { ((element.split(" ")).head).toLong } }
package flink.streaming import java.util.Properties import org.apache.flink.streaming.api.scala._ import org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer import org.apache.flink.api.common.serialization.SimpleStringSchema import org.apache.flink.streaming.api.windowing.time.Time import org.apache.flink.streaming.api.TimeCharacteristic import org.apache.flink.streaming.api.functions.sink.RichSinkFunction import org.apache.flink.streaming.api.CheckpointingMode import org.apache.flink.streaming.api.functions.AssignerWithPeriodicWatermarks object StreamWithEventTimeAndWaterMarks { def main(args: Array[String]): Unit = { val kafkaProps = new Properties() //kafka的一些属性 kafkaProps.setProperty("bootstrap.servers", "bigdata01:9092") //所在的消费组 kafkaProps.setProperty("group.id", "group2") //获取当前的执行环境 val evn = StreamExecutionEnvironment.getExecutionEnvironment //配制处理数据的时候使用event time evn.setStreamTimeCharacteristic(TimeCharacteristic.EventTime) //kafka的consumer,test1是要消费的topic val kafkaSource = new FlinkKafkaConsumer[String]("event_time",new SimpleStringSchema,kafkaProps) //添加自定义的 TimestampsAndWatermarks kafkaSource.assignTimestampsAndWatermarks(new CustomWaterMarks) //设置从最新的offset开始消费 //kafkaSource.setStartFromGroupOffsets() kafkaSource.setStartFromLatest() //自动提交offset kafkaSource.setCommitOffsetsOnCheckpoints(true) //flink的checkpoint的时间间隔 //evn.enableCheckpointing(2000) //添加consumer val stream = evn.addSource(kafkaSource) evn.enableCheckpointing(2000, CheckpointingMode.EXACTLY_ONCE) //stream.setParallelism(3) val text = stream.flatMap{ _.toLowerCase().split(" ").drop(1).filter { _.nonEmpty} } .map{(_,1)} .keyBy(0) .timeWindow(Time.seconds(5)) .sum(1) .map(x=>{(x._1,(new Integer(x._2)))}) text.print() //启动执行 //text.addSink(new Ssinks()) evn.execute("kafkawd") } }
打印结果 partition-2中的数据因为超时没有出现 1> (big,14) 4> (is,14) 1> (+0,1) 2> (+1,1) 3> (partition-1--this,7) 4> (+15,1) 3> (+12,1) 1> (partition-0--this,7) 3> (+6,1) 1> (+16,1) 4> (+10,1) 2> (+18,1) 4> (+7,1) 3> (+3,1) 2> (+9,1) 3> (+19,1) 2> (+13,1) 3> (a,14) 2> (+4,1)
来自 “ ITPUB博客 ” ,链接:http://blog.itpub.net/31506529/viewspace-2637182/,如需转载,请注明出处,否则将追究法律责任。
转载于:http://blog.itpub.net/31506529/viewspace-2637182/