技术库 > 网站架构

Hadoop的DistributedCache

技术库:tec.5lulu.com

DistributedCache是Hadoop的一个分布式文件缓存类,使用它有时候能完成一些比较方便的事,DistributedCache第一个比较方便的作用就是来完成分布式文件共享这件事,第二个比较有用的场景,就是在执行一些join操作时,将小表放入cache中,来提高连接效率。 

那么,散仙今天要介绍的是如何使用DistributedCache来共享全局的缓存文件。 
下面我们先通过一个表格来看下,在hadoop中,使用全局变量或全局文件共享的几种方法 
序号 方法
1 使用Configuration的set方法,只适合数据内容比较小的场景
2 将共享文件放在HDFS上,每次都去读取,效率比较低
3 将共享文件放在DistributedCache里,在setup初始化一次后,即可多次使用,缺点是不支持修改操作,仅能读取


通过DistributedCache的addCacheFile方法,我们把HDFS上的一些文件,在hadoop任务启动时,就载入缓存里面,以供全局使用,使用这个方法时,我们需要注意几点,首先我们的本地文件,需要上传到HDFS上,然后再这个方法里,写入加载路径,接下来,我们就可以,在setup初始化时,读取出,其内容,然后在map或reduce方法,执行时,就可以实时的使用这个文件的一些内容了。 
散仙,测试共享的一个文件内容如下: 


Hadoop的DistributedCache,by 5lulu.com
代码如下,注意散仙在setup方法里,读取了文件内容,并打印:  
package com.qin.testdistributed;  
  
import java.io.File;  
import java.io.FileReader;  
import java.io.IOException;  
import java.net.URI;  
import java.util.Scanner;  
  
import org.apache.hadoop.conf.Configuration;  
import org.apache.hadoop.filecache.DistributedCache;  
import org.apache.hadoop.fs.FSDataInputStream;  
import org.apache.hadoop.fs.FileSystem;  
import org.apache.hadoop.fs.Path;  
import org.apache.hadoop.io.IntWritable;  
import org.apache.hadoop.io.LongWritable;  
import org.apache.hadoop.io.Text;  
import org.apache.hadoop.mapred.JobConf;  
import org.apache.hadoop.mapreduce.Job;  
import org.apache.hadoop.mapreduce.Mapper;  
import org.apache.hadoop.mapreduce.Reducer;  
import org.apache.hadoop.mapreduce.lib.db.DBConfiguration;  
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;  
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;  
import org.apache.log4j.pattern.LogEvent;  
   
import org.slf4j.Logger;  
import org.slf4j.LoggerFactory;  
  
import com.qin.operadb.WriteMapDB;  
   
  
/** 
 * 测试hadoop的全局共享文件 
 * 使用DistributedCached 
 *  
 * 大数据技术交流群: 37693216 
 * @author qindongliang 
 *  
 * ***/  
public class TestDistributed {   
    private static Logger logger=LoggerFactory.getLogger(TestDistributed.class);   
      
    private static class FileMapper extends Mapper<LongWritable, Text, Text, IntWritable>{  
          
            Path path[]=null;  
              
        /** 
         * Map函数前调用 
         *  
         * */  
        @Override  
        protected void setup(Context context)  
                throws IOException, InterruptedException {  
          logger.info("开始启动setup了哈哈哈哈");  
            // System.out.println("运行了.........");  
          Configuration conf=context.getConfiguration();  
           path=DistributedCache.getLocalCacheFiles(conf);  
           System.out.println("获取的路径是:  "+path[0].toString());  
         //  FileSystem fs = FileSystem.get(conf);  
           FileSystem fsopen= FileSystem.getLocal(conf);  
         FSDataInputStream in = fsopen.open(path[0]);  
          // System.out.println(in.readLine());  
//         for(Path tmpRefPath : path) {  
//             if(tmpRefPath.toString().indexOf("ref.png") != -1) {  
//                 in = reffs.open(tmpRefPath);  
//                 break;  
//             }  
//         }  
             
     // FileReader reader=new FileReader("file://"+path[0].toString());  
//      File f=new File("file://"+path[0].toString());  
      // FSDataInputStream in=fs.open(new Path(path[0].toString()));  
         Scanner scan=new Scanner(in);  
           while(scan.hasNext()){  
               System.out.println(Thread.currentThread().getName()+"扫描的内容:  "+scan.next());  
           }  
           scan.close();  
//        
        // System.out.println("size: "+path.length);  
              
              
        }  
          
          
        @Override  
        protected void map(LongWritable key, Text value,Context context)  
                throws IOException, InterruptedException {  
           
        //  System.out.println("map    aaa");  
            //logger.info("Map里的任务");  
            System.out.println("map里输出了");  
        //  logger.info();  
            context.write(new Text(""), new IntWritable(0));    
        }   
          
         @Override  
        protected void cleanup(Context context)  
                throws IOException, InterruptedException {                 
             logger.info("清空任务了。。。。。。");  
        }            
    }        
      
    private static class  FileReduce extends Reducer<Object, Object, Object, Object>{  
                    
        @Override  
        protected void reduce(Object arg0, Iterable<Object> arg1,  
                 Context arg2)throws IOException, InterruptedException {   
            System.out.println("我是reduce里面的东西");  
        }  
    }  
      
    public static void main(String[] args)throws Exception {  
          
        JobConf conf=new JobConf(TestDistributed.class);  
        //conf.set("mapred.local.dir", "/root/hadoop");  
         //Configuration conf=new Configuration();  
          
        // conf.set("mapred.job.tracker","192.168.75.130:9001");  
        //读取person中的数据字段  
           //conf.setJar("tt.jar");  
           
        //注意这行代码放在最前面,进行初始化,否则会报  
         String inputPath="hdfs://192.168.75.130:9000/root/input";        
         String outputPath="hdfs://192.168.75.130:9000/root/outputsort";  
           
        Job job=new Job(conf, "a");  
        DistributedCache.addCacheFile(new URI("hdfs://192.168.75.130:9000/root/input/f1.txt"), job.getConfiguration());  
        job.setJarByClass(TestDistributed.class);  
        System.out.println("运行模式:  "+conf.get("mapred.job.tracker"));  
        /**设置输出表的的信息  第一个参数是job任务,第二个参数是表名,第三个参数字段项**/  
       FileSystem fs=FileSystem.get(job.getConfiguration());  
          
          Path pout=new Path(outputPath);  
          if(fs.exists(pout)){  
              fs.delete(pout, true);  
              System.out.println("存在此路径, 已经删除......");  
          }   
         /**设置Map类**/  
        // job.setOutputKeyClass(Text.class);  
         //job.setOutputKeyClass(IntWritable.class);  
          job.setMapOutputKeyClass(Text.class);  
          job.setMapOutputValueClass(IntWritable.class);  
         job.setMapperClass(FileMapper.class);  
         job.setReducerClass(FileReduce.class);  
         FileInputFormat.setInputPaths(job, new Path(inputPath));  //输入路径  
         FileOutputFormat.setOutputPath(job, new Path(outputPath));//输出路径              
        System.exit(job.waitForCompletion(true) ? 0 : 1);           
    }   
}  
在web页面上,查询日志输入情况,如下截图所示 

Hadoop的DistributedCache,by 5lulu.com
当然,我们也可以根据web上的路径,到对应的日志目录下,查找日志的内容,看到上图就说明,我们上传的共享文件,读取成功了,只要在setup方法里面进行初始化后,对我们的程序来说,就是全局共享了,然后我们就可以结合我们的业务逻辑,来处理一些事了。 

最后,在简单总结一下:DistributedCache文件共享的模式,只能在集群的环境中使用,如果在Local模式下测试,就会报如下的文件找不到异常:    
运行模式:  local  
存在此路径, 已经删除......  
WARN - NativeCodeLoader.<clinit>(52) | Unable to load native-hadoop library for your platform... using builtin-java classes where applicable  
WARN - JobClient.copyAndConfigureFiles(746) | Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same.  
WARN - JobClient.copyAndConfigureFiles(870) | No job jar file set.  User classes may not be found. See JobConf(Class) or JobConf#setJar(String).  
INFO - FileInputFormat.listStatus(237) | Total input paths to process : 1  
WARN - LoadSnappy.<clinit>(46) | Snappy native library not loaded  
INFO - TrackerDistributedCacheManager.downloadCacheObject(423) | Creating f1.txt in /root/hadoop1.2/hadooptmp/mapred/local/archive/-4778653900406898379_1788685676_88844454/192.168.75.130/root/input-work--1953076903080970848 with rwxr-xr-x  
INFO - TrackerDistributedCacheManager.downloadCacheObject(463) | Cached hdfs://192.168.75.130:9000/root/input/f1.txt as /root/hadoop1.2/hadooptmp/mapred/local/archive/-4778653900406898379_1788685676_88844454/192.168.75.130/root/input/f1.txt  
INFO - TrackerDistributedCacheManager.localizePublicCacheObject(486) | Cached hdfs://192.168.75.130:9000/root/input/f1.txt as /root/hadoop1.2/hadooptmp/mapred/local/archive/-4778653900406898379_1788685676_88844454/192.168.75.130/root/input/f1.txt  
INFO - JobClient.monitorAndPrintJob(1380) | Running job: job_local697121855_0001  
INFO - LocalJobRunner$Job.run(340) | Waiting for map tasks  
INFO - LocalJobRunner$Job$MapTaskRunnable.run(204) | Starting task: attempt_local697121855_0001_m_000000_0  
INFO - Task.initialize(534) |  Using ResourceCalculatorPlugin : null  
INFO - MapTask.runNewMapper(729) | Processing split: hdfs://192.168.75.130:9000/root/input/f1.txt:0+31  
INFO - MapTask$MapOutputBuffer.<init>(949) | io.sort.mb = 100  
INFO - MapTask$MapOutputBuffer.<init>(961) | data buffer = 79691776/99614720  
INFO - MapTask$MapOutputBuffer.<init>(962) | record buffer = 262144/327680  
INFO - TestDistributed$FileMapper.setup(60) | 开始启动setup了哈哈哈哈  
获取的路径是:  /root/hadoop1.2/hadooptmp/mapred/local/archive/-4778653900406898379_1788685676_88844454/192.168.75.130/root/input/f1.txt  
INFO - MapTask$MapOutputBuffer.flush(1289) | Starting flush of map output  
INFO - LocalJobRunner$Job.run(348) | Map task executor complete.  
WARN - LocalJobRunner$Job.run(435) | job_local697121855_0001  
java.lang.Exception: java.io.FileNotFoundException: File /root/hadoop1.2/hadooptmp/mapred/local/archive/-4778653900406898379_1788685676_88844454/192.168.75.130/root/input/f1.txt does not exist.  
    at org.apache.hadoop.mapred.LocalJobRunner$Job.run(LocalJobRunner.java:354)  
Caused by: java.io.FileNotFoundException: File /root/hadoop1.2/hadooptmp/mapred/local/archive/-4778653900406898379_1788685676_88844454/192.168.75.130/root/input/f1.txt does not exist.  
    at org.apache.hadoop.fs.RawLocalFileSystem.getFileStatus(RawLocalFileSystem.java:402)  
    at org.apache.hadoop.fs.FilterFileSystem.getFileStatus(FilterFileSystem.java:255)  
    at org.apache.hadoop.fs.ChecksumFileSystem$ChecksumFSInputChecker.<init>(ChecksumFileSystem.java:125)  
    at org.apache.hadoop.fs.ChecksumFileSystem.open(ChecksumFileSystem.java:283)  
    at org.apache.hadoop.fs.FileSystem.open(FileSystem.java:427)  
    at com.qin.testdistributed.TestDistributed$FileMapper.setup(TestDistributed.java:67)  
    at org.apache.hadoop.mapreduce.Mapper.run(Mapper.java:142)  
    at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:764)  
    at org.apache.hadoop.mapred.MapTask.run(MapTask.java:364)  
    at org.apache.hadoop.mapred.LocalJobRunner$Job$MapTaskRunnable.run(LocalJobRunner.java:223)  
    at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)  
    at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:334)  
    at java.util.concurrent.FutureTask.run(FutureTask.java:166)  
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110)  
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603)  
    at java.lang.Thread.run(Thread.java:722)  
INFO - JobClient.monitorAndPrintJob(1393) |  map 0% reduce 0%  
INFO - JobClient.monitorAndPrintJob(1448) | Job complete: job_local697121855_0001  
INFO - Counters.log(585) | Counters: 0  

如果你很幸运,在1.x的hadoop里看到如下所示的异常,那么你应该考虑如下的几个问题,第一,是不是以Local模式启动的MR任务,第二读取时的路径是不是有问题,使用DistributedCache共享的文件,会在我们每个节点上配置的目录里面找到对应的共享文件:  
<?xml version="1.0"?>  
<?xml-stylesheet type="text/xsl" href="configuration.xsl"?>  
  
<!-- Put site-specific property overrides in this file. -->  
  
<configuration>  
<!-- jobtracker的master地址-->  
<property>   
<name>mapred.job.tracker</name>   
<value>192.168.75.130:9001</value>   
</property>  
<property>  
<!-- hadoop的日志输出指定目录-->  
<name>mapred.local.dir</name>  
<value>/root/hadoop1.2/mylogs</value>  
</property>  
  
</configuration>   

共享的文件,会被下载到每个节点上的指定的文件夹里找到。 
散仙找的一个的路径: 
/root/hadoop1.2/mylogs/taskTracker/distcache/2726204645197711229_1788685676_88844454/192.168.75.130/root/input
其他的节点上也一样,只不过IP地址不一样,截图如下 

Hadoop的DistributedCache,by 5lulu.com 
至此,我们就可以使用轻松的来使用DistributedCache来共享一些比较大的文件,或压缩包了。

Hadoop的DistributedCache


本文链接 http://tec.5lulu.com/detail/105dgn2i5p6wm8sfb.html

我来评分 :6.1
0

转载注明:转自5lulu技术库

本站遵循:署名-非商业性使用-禁止演绎 3.0 共享协议

www.5lulu.com