Hadoop MapFile reader doesn't detect a file in distributed Cache

Question

I need help regarding mapfile reader.
I add the files into the cache by -files options

yarn jar HadoopProjects.jar rsProject.driver -files hdfs://localhost:8020/data/mapFileTestFolder.tar.gz....

here I call it

@SuppressWarnings("deprecation")
@Override
protected void setup(Context context) {
    try {
        Path[] cacheLocalFiles = DistributedCache.getLocalCacheFiles(context.getConfiguration());
        logF.info("reducer started setup");

            for (Path path:cacheLocalFiles) {
                logF("reducer setup " + path.getName().toString());
                if (path.getName().toString().contains("mapFileTestFolder.tar.gz")) {
                    URI mapUri = new File(path.toString() + "/mapFileTestFolder").toURI();
                    logF.info("depReader init begins URI = " + mapUri.toString());
                    depReader = new MapFile.Reader(FileSystem.get(context.getConfiguration()),mapUri.toString(), context.getConfiguration());
                    logF.info("depReader init ends");

                }
            }
        } catch (IOException e) {
            e.printStackTrace();
            logF.info("depReader init error - " + e);
        } 
        //some other lines

}

Here what I see in logs

2014-03-11 08:31:09,305 INFO [main] rsProject.myReducer: depReader init begins URI = file:/home/hadoop/Training/hadoop_work/mapred/nodemanager/usercache/hadoop/appcache/application_1394318775013_0079/container_1394318775013_0079_01_000005/mapFileTestFolder.tar.gz/mapFileTestFolder

2014-03-11 08:31:09,345 INFO [main] rsProject.myReducer: depReader init error - java.io.FileNotFoundException: File file:/home/hadoop/Training/hadoop_work/mapred/nodemanager/usercache/hadoop/appcache/application_1394318775013_0079/container_1394318775013_0079_01_000005/mapFileTestFolder.tar.gz/mapFileTestFolder/data does not exist

mapFileTestFolder.tar.gz - this is a compressed map file file (with index and data in it)

I guess this file exists in the distributed cache as the runner goes into the condition if the same matches.
Why does this happen? =/

Any help is appreciated
Thanks

Alexander Komarov Alexander Komarov · Accepted Answer · 2014-03-13T10:24:12

the question is resolved. My stupid fault =/ I should have used the command adding an archive to the distributed cache, not as a file.

Hadoop MapFile reader doesn't detect a file in distributed Cache

1 Answers