java - 我怎样才能不间断地运行 Spark 流

Question

我正在尝试在 twitter-streaming 的帮助下保存来自 twitter 的推文。但我有一个问题:我的程序在一段时间后停止工作(取决于接近 4-5 秒的 1 毫秒的批处理间隔)。那么，你能帮我解决这个问题吗)。请告诉我出了什么问题？

当批处理间隔接近 100 毫秒时，我会看到一些记录，例如

19/08/06 23:45:26 INFO BlockRDD: Removing RDD 103 from persistence list
19/08/06 23:45:26 INFO BlockManager: Removing RDD 103
19/08/06 23:45:26 INFO TwitterInputDStream: Removing blocks of RDD BlockRDD[103] at createStream at Twitter.java:35 of time 1565124324340 ms
19/08/06 23:45:26 INFO ReceivedBlockTracker: Deleting batches: 1565124324320 ms
19/08/06 23:45:26 INFO InputInfoTracker: remove old batch metadata: 1565124324320 ms
-------------------------------------------
Time: 1565124325500 ms

当批处理间隔“大”并且任何数据不可用时，我只会看到有关 Spark UI 启动和完成的消息。

package TwitterAnalysis;

import org.apache.spark.*;
import org.apache.spark.storage.StorageLevel;
import org.apache.spark.streaming.*;
import org.apache.spark.streaming.api.java.*;
import org.apache.spark.streaming.twitter.*;

import twitter4j.Status;



public class Twitter {

    private static void setTwitterOAuth() {
        System.setProperty("twitter4j.oauth.consumerKey", TwitterOAuthKey.consumerKey);
        System.setProperty("twitter4j.oauth.consumerSecret", TwitterOAuthKey.consumerSecret);
        System.setProperty("twitter4j.oauth.accessToken", TwitterOAuthKey.accessToken);
        System.setProperty("twitter4j.oauth.accessTokenSecret", TwitterOAuthKey.accessTokenSecret);
    }



    public static void main(String [] args) {

        setTwitterOAuth();

        SparkConf conf = new SparkConf().setMaster("local[*]")
                                         .setAppName("SparkTwitter");

      //  JavaSparkContext sparkContext = new JavaSparkContext(conf);
        JavaStreamingContext jssc = new JavaStreamingContext(conf, new Duration(10000));


        JavaReceiverInputDStream<Status> twitterStream = TwitterUtils.createStream(jssc);

        //Stream that contains just tweets in english
        JavaDStream<Status> enTweetsDStream=twitterStream.filter((status) -> "en".equalsIgnoreCase(status.getLang()));
        enTweetsDStream.persist(StorageLevel.MEMORY_AND_DISK());


        enTweetsDStream.print();
        jssc.start();


    }

}

Answer 1

根据这个答案:Spark 2.0.0 twitter streaming driver is no longer available Spark 2.0及更高版本没有可用的twitter-streaming-driver。解决方案选择早期版本的Spark)