Question

它显示它创建了缓存文件。但是，当我去查看文件不存在的位置时，当我尝试从我的映射器中读取时，它显示文件未找到异常。

这是我尝试运行的代码：

    JobConf conf2 = new JobConf(getConf(), CorpusCalculator.class);
    conf2.setJobName("CorpusCalculator2");

    //Distributed Caching of the file emitted by the reducer2 is done here
    conf2.addResource(new Path("/opt/hadoop1/conf/core-site.xml"));
    conf2.addResource(new Path("/opt/hadoop1/conf/hdfs-site.xml"));

    //cacheFile(conf2, new Path(outputPathofReducer2));

    conf2.setNumReduceTasks(1);
    //conf2.setOutputKeyComparatorClass()

    conf2.setMapOutputKeyClass(FloatWritable.class);
    conf2.setMapOutputValueClass(Text.class);


    conf2.setOutputKeyClass(Text.class);
    conf2.setOutputValueClass(Text.class);

    conf2.setMapperClass(MapClass2.class);
    conf2.setReducerClass(Reduce2.class);



    FileInputFormat.setInputPaths(conf2, new Path(inputPathForMapper1));
    FileOutputFormat.setOutputPath(conf2, new Path(outputPathofReducer3));

    DistributedCache.addCacheFile(new Path("/sunilFiles/M51.txt").toUri(),conf2);
    JobClient.runJob(conf

日志：

13/04/27 04:43:40 INFO filecache.TrackerDistributedCacheManager: Creating M51.txt in /tmp1/mapred/local/archive/-1731849462204707023_-2090562221_1263420527/localhost/sunilFiles-work-2204204368663038938 with rwxr-xr-x

13/04/27 04:43:40 INFO filecache.TrackerDistributedCacheManager: Cached /sunilFiles/M51.txt as /tmp1/mapred/local/archive/-1731849462204707023_-2090562221_1263420527/localhost/sunilFiles/M51.txt

13/04/27 04:43:40 INFO filecache.TrackerDistributedCacheManager: Cached /sunilFiles/M51.txt as /tmp1/mapred/local/archive/-1731849462204707023_-2090562221_1263420527/localhost/sunilFiles/M51.txt

13/04/27 04:43:40 INFO mapred.JobClient: Running job: job_local_0003

13/04/27 04:43:40 INFO mapred.Task:  Using ResourceCalculatorPlugin : o
org.apache.hadoop.util.LinuxResourceCalculatorPlugin@8c2df1

13/04/27 04:43:40 INFO mapred.MapTask: numReduceTasks: 1

13/04/27 04:43:40 INFO mapred.MapTask: io.sort.mb = 100

13/04/27 04:43:40 INFO mapred.MapTask: data buffer = 79691776/99614720

13/04/27 04:43:40 INFO mapred.MapTask: record buffer = 262144/327680

在configure()内：

Exception reading DistribtuedCache: java.io.FileNotFoundException: /tmp1/mapred/local/archive/-1731849462204707023_-2090562221_1263420527/localhost/sunilFiles/M51.txt (Is a directory)

Inside setup(): /tmp1/mapred/local/archive/-1731849462204707023_-2090562221_1263420527/localhost/sunilFiles/M51.txt

13/04/27 04:43:41 WARN mapred.LocalJobRunner: job_local_0003

请帮助我，我一直在寻找解决方案，持续6个小时，明天我有一份作业提交。非常感谢你。

Answer 1

你可能想尝试更简单的-files选项。为了能够使用它，驱动程序类需要扩展Configured并实现Tool。

例如，

hadoop jar jarname.jar driverclass -files file1.xml，file2.txt

在mapper或reducer中：

BufferedReader reader1 = new BufferedReader(new FileReader("file1.xml"));
BufferedReader reader2 = new BufferedReader(new FileReader("file2.txt"));

Answer 2

我通过使用copyMerge（）属性解决了这个问题，该属性将各种机器中存在的所有文件合并到一个文件中，并且我成功地使用了该文件。如果我使用普通文件则失败。谢谢你们的回复。

Hadoop文件中的分布式缓存未找到异常

2 个答案: