在线实时大数据平台Storm输入源共享试验

1、背景：topology程序提交集群模式运行试验，验证在同一文件输入源情况下，worker之间是否会重复输入处理，以及数据变量能否在不同worker之间共享，如果文件新增数据，topology会不会获取最新数据处理。

2、试验代码：

package cn.wc;import org.apache.storm.Config;import org.apache.storm.StormSubmitter;import org.apache.storm.generated.AlreadyAliveException;import org.apache.storm.generated.AuthorizationException;import org.apache.storm.generated.InvalidTopologyException;import org.apache.storm.topology.TopologyBuilder;import org.apache.storm.tuple.Fields;public class TopologyMain {public static void main(String[] args) throws InterruptedException {         //ConfigurationConfig conf = new Config();conf.setNumWorkers(3);//设置3个进程conf.put("wordsFile", args[0]);conf.put("output", args[1]);        //Topology definitionTopologyBuilder builder = new TopologyBuilder();builder.setSpout("word-reader",new WordReader(),3);builder.setBolt("word-normalizer",new WordNormalizer(),3).setNumTasks(6).shuffleGrouping("word-reader");builder.setBolt("word-counter", new WordCounter(),3).fieldsGrouping("word-normalizer", new Fields("word"));//集群模式try {StormSubmitter.submitTopology("Getting-Started-Toplogie", conf, builder.createTopology());} catch (AlreadyAliveException e) {// TODO Auto-generated catch blocke.printStackTrace();} catch (InvalidTopologyException e) {// TODO Auto-generated catch blocke.printStackTrace();} catch (AuthorizationException e) {// TODO Auto-generated catch blocke.printStackTrace();}  }}

package cn.wc;import java.io.BufferedReader;import java.io.FileNotFoundException;import java.io.FileReader;import java.util.Map;import org.apache.storm.spout.SpoutOutputCollector;import org.apache.storm.task.TopologyContext;import org.apache.storm.topology.OutputFieldsDeclarer;import org.apache.storm.topology.base.BaseRichSpout;import org.apache.storm.tuple.Fields;import org.apache.storm.tuple.Values;//Spout作为数据源，它实现了IRichSpout接口，功能是读取一个文本文件并把它的每一行内容发送给bolt。public class WordReader extends BaseRichSpout {private SpoutOutputCollector collector;private FileReader fileReader;private boolean completed = false;public void ack(Object msgId) {System.out.println("OK:"+msgId);}public void close() {}public void fail(Object msgId) {System.out.println("FAIL:"+msgId);}/** * The only thing that the methods will do It is emit each  * file line * spout最主要的方法，读取文本文件，并把它的每一行发射出去（给bolt）      * 这个方法会不断被调用，为了降低它对CPU的消耗，当任务完成时让它sleep一下  */public void nextTuple() {/** * The nextuple it is called forever, so if we have been readed the file * we will wait and then return */if(completed){try {Thread.sleep(1000);} catch (InterruptedException e) {//Do nothing}return;}String str;//Open the readerBufferedReader reader = new BufferedReader(fileReader);try{//Read all lineswhile((str = reader.readLine()) != null){/** * By each line emmit a new value with the line as a their * 发射每一行，Values是一个ArrayList的实现  */this.collector.emit(new Values(str),str);}}catch(Exception e){throw new RuntimeException("Error reading tuple",e);}finally{completed = true;}}/** * We will create the file and get the collector object * 三个参数，第一个是创建Topology时的配置，第二个是所有的Topology数据，第三个是用来把Spout的数据发射给bolt      *   */public void open(Map conf, TopologyContext context,SpoutOutputCollector collector) {try {//获取创建Topology时指定的要读取的文件路径  this.fileReader = new FileReader(conf.get("wordsFile").toString());} catch (FileNotFoundException e) {throw new RuntimeException("Error reading file ["+conf.get("wordFile")+"]");}//初始化发射器this.collector = collector;}/** * Declare the output field "word" */public void declareOutputFields(OutputFieldsDeclarer declarer) {declarer.declare(new Fields("line"));}}

package cn.wc;import org.apache.storm.topology.BasicOutputCollector;import org.apache.storm.topology.OutputFieldsDeclarer;import org.apache.storm.topology.base.BaseBasicBolt;import org.apache.storm.tuple.Fields;import org.apache.storm.tuple.Tuple;import org.apache.storm.tuple.Values;//Spout已经成功读取文件并把每一行作为一个tuple(在Storm数据以tuple的形式传递)发射过来，我们这里需要创建两个bolt分别来负责解析每一行和对单词计数。//Bolt中最重要的是execute方法，每当一个tuple传过来时它便会被调用。public class WordNormalizer extends BaseBasicBolt {public void cleanup() {}/** * The bolt will receive the line from the * words file and process it to Normalize this line *  * The normalize will be put the words in lower case * and split the line to get all words in this  * bolt中最重要的方法，每当接收到一个tuple时，此方法便被调用      * 这个方法的作用就是把文本文件中的每一行切分成一个个单词，并把这些单词发射出去（给下一个bolt处理）  */public void execute(Tuple input, BasicOutputCollector collector) {        String sentence = input.getString(0);        String[] words = sentence.split(" ");        for(String word : words){            word = word.trim();            if(!word.isEmpty()){                word = word.toLowerCase();                collector.emit(new Values(word));            }        }}/** * The bolt will only emit the field "word"  */public void declareOutputFields(OutputFieldsDeclarer declarer) {declarer.declare(new Fields("word"));}}

package cn.wc;import java.io.File;import java.io.FileOutputStream;import java.io.IOException;import java.util.HashMap;import java.util.Map;import org.apache.storm.task.TopologyContext;import org.apache.storm.topology.BasicOutputCollector;import org.apache.storm.topology.OutputFieldsDeclarer;import org.apache.storm.topology.base.BaseBasicBolt;import org.apache.storm.tuple.Tuple;public class WordCounter extends BaseBasicBolt {Integer id;String name;Map<String, Integer> counters;String output=null;/** * At the end of the spout (when the cluster is shutdown * We will show the word counters * Topology执行完毕的清理工作，比如关闭连接、释放资源等操作都会写在这里  */@Overridepublic void cleanup() {/*System.out.println("-- Word Counter ["+name+"-"+id+"] --");for(Map.Entry<String, Integer> entry : counters.entrySet()){System.out.println(entry.getKey()+": "+entry.getValue());}*/}/** * On create  */@Overridepublic void prepare(Map stormConf, TopologyContext context) {this.counters = new HashMap<String, Integer>();this.name = context.getThisComponentId();this.id = context.getThisTaskId();output=stormConf.get("output").toString();}@Overridepublic void declareOutputFields(OutputFieldsDeclarer declarer) {}@Overridepublic void execute(Tuple input, BasicOutputCollector collector) {String str = input.getString(0);/** * If the word dosn't exist in the map we will create * this, if not We will add 1  */if(!counters.containsKey(str)){counters.put(str, 1);}else{Integer c = counters.get(str) + 1;counters.put(str, c);}//写入文件try{File file=new File(output);        if(!file.exists())            file.createNewFile();        FileOutputStream out=new FileOutputStream(file,true);                for(Map.Entry<String, Integer> entry : counters.entrySet()){            StringBuffer sb=new StringBuffer();            sb.append(entry.getKey()+": "+entry.getValue());            sb.append("\r\n");            out.write(sb.toString().getBytes("utf-8"));        }  }catch (IOException e){e.printStackTrace();}      }}

3、结果分析：

集群环境下执行：storm jar /mnt/wc.jar cn.wc.TopologyMain /mnt/words.txt /tmp/topo.log
/*并行和通信试验：
 * 设置worker为3，启动3个进程来服务这个topology
 * spout/bolt的线程线程设置为3，默认对应一个task，就是一个进程跑一个task，总共有9个；
 * 现在对word-normalizer这个bolt设置任务6个，那就是每个进程分2个，现在总共12个task；
 * 总的来说：worker进程有3个，executor线程有9个，task任务有12个；
 * 输入：/mnt/words.txt 输出：/tmp/topo.log
*/

1）storm list发现task是15个，不是12个，怎么算就有点疑惑了；

2）输入的词汇，明显被重复统计3次，也就是说3个executor在同一文件输入源下，不会自动去协调输入记录从而排斥；

3）topology程序中设置的变量，无法再executor之间共享；

4）输入的文件新增词汇，topology没有及时去获取统计，当然topology仍然在集群中运行

4、总结：

      1）一个topology被提交到不同节点的不同worker（进程）分布执行，要按照独立进程来看；

      2）worker内要有自己唯一的输入源，同时要确保输入源是持续提供；

      3）要在worker之间共享数据变量，只能通过其他办法，如redis来存储；

      也就是说：topology被提交到集群分布式执行，不同worker之间是独立进程运作。