为什么Spark流媒体不会从Kafka主题中读取

Question

• 火花流 1.6.0
• apache kafka 10.0.1

我使用火花流从sample主题中读取。该代码在没有错误或异常的情况下运行，但我没有通过print()方法在控制台上获得任何数据。

我检查了主题中是否有消息：

./bin/kafka-console-consumer.sh 
    --zookeeper ip-172-xx-xx-xxx:2181 
    --topic sample 
    --from-beginning

我收到消息：

message no. 1
message no. 2
message no. 3
message no. 4
message no. 5

运行流式作业的命令：

./bin/spark-submit 
    --conf "spark.executor.extraJavaOptions=-XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:MaxDirectMemorySize=512m" 
    --jars /home/ubuntu/zifferlabs/target/ZifferLabs-1-jar-with-dependencies.jar 
    --class "com.zifferlabs.stream.SampleStream" 
    /home/ubuntu/zifferlabs/src/main/java/com/zifferlabs/stream/SampleStream.java

这是整个代码：

import java.util.Arrays;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Map;
import java.util.Set;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.function.Function;
import org.apache.spark.streaming.Duration;
import org.apache.spark.streaming.api.java.JavaDStream;
import org.apache.spark.streaming.api.java.JavaPairInputDStream;
import org.apache.spark.streaming.api.java.JavaStreamingContext;
import org.apache.spark.streaming.kafka.KafkaUtils;
import kafka.serializer.DefaultDecoder;
import kafka.serializer.StringDecoder;
import scala.Tuple2;
public class SampleStream {
  private static void processStream() {
    SparkConf conf = new SparkConf().setAppName("sampleStream")
            .setMaster("local[3]")
            .set("spark.serializer", "org.apache.spark.serializer.JavaSerializer")
            .set("spark.driver.memory", "2g").set("spark.streaming.blockInterval", "1000")
            .set("spark.driver.allowMultipleContexts", "true")
            .set("spark.scheduler.mode", "FAIR");
    JavaStreamingContext jsc = new JavaStreamingContext(conf, new Duration(Long.parseLong("2000")));
    String[] topics = "sample".split(",");
    Set<String> topicSet = new HashSet<String>(Arrays.asList(topics));
    Map<String, String> props = new HashMap<String, String>();
    props.put("metadata.broker.list", "ip-172-xx-xx-xxx:9092");
    props.put("kafka.consumer.id", "sample_con");
    props.put("group.id", "sample_group");
    props.put("zookeeper.connect", "ip-172-xx-xx-xxx:2181");
    props.put("zookeeper.connection.timeout.ms", "16000");
    JavaPairInputDStream<String, byte[]> kafkaStream =
      KafkaUtils.createDirectStream(jsc, String.class, byte[].class, StringDecoder.class,
                                    DefaultDecoder.class, props, topicSet);
    JavaDStream<String> data = kafkaStream.map(new Function<Tuple2<String,byte[]>, String>() {
      public String call(Tuple2<String, byte[]> arg0) throws Exception {
        System.out.println("$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ value is: " + arg0._2().toString());
        return arg0._2().toString();
      }
    });
    data.print();
    System.out.println("Spark Streaming started....");
    jsc.checkpoint("/home/spark/sparkChkPoint");
    jsc.start();
    jsc.awaitTermination();
    System.out.println("Stopped Spark Streaming");
  }
  public static void main(String[] args) {
    processStream();
  }
}

Answer 1

i think think 你有正确的代码，但是执行的命令行是不正确的。

您的spark-submit应用程序如下(为简单起见，删除了格式化的矿山 spark.executor.extraJavaOptions(：

./bin/spark-submit 
  --jars /home/ubuntu/zifferlabs/target/ZifferLabs-1-jar-with-dependencies.jar 
  --class "com.zifferlabs.stream.SampleStream" 
  /home/ubuntu/zifferlabs/src/main/java/com/zifferlabs/stream/SampleStream.java

i think 它将无法使用，因为spark-submit提交了您的Java 源代码而不是可执行代码。

请 spark-submit您的申请如下：

./bin/spark-submit 
  --class "com.zifferlabs.stream.SampleStream" 
  /home/ubuntu/zifferlabs/target/ZifferLabs-1-jar-with-dependencies.jar

是--class，可以定义您的Spark应用程序的"入口点"和带有依赖项的代码(作为spark-submit的唯一输入参数(。

给它一个镜头并报告！