【hadoop】 3003-mapreduce任务的提交

1、处理数据的作业达成jar包并上传hdfs

[hadoop@cloud01 HDFSdemo]$ pwd
/home/hadoop/workspace/HDFSdemo
[hadoop@cloud01 HDFSdemo]$ ll
total 139844
drwxrwxr-x. 5 hadoop hadoop      4096 Feb 24 18:10 bin
-rw-rw-r--. 1 hadoop hadoop       440 Feb 20 06:56 core-site.xml
-rw-rw-r--. 1 hadoop hadoop       256 Feb 20 06:56 hdfs-site.xml
drwxrwxr-x. 2 hadoop hadoop      4096 Feb 20 06:34 lib
-rw-rw-r--. 1 hadoop hadoop       253 Feb 20 06:56 mapred-site.xml
drwxrwxr-x. 5 hadoop hadoop      4096 Feb 24 18:10 src
-rw-rw-r--. 1 hadoop hadoop 143167974 Feb 24 21:41  wc.jar
-rw-rw-r--. 1 hadoop hadoop       434 Feb 20 06:56 yarn-site.xml

2、启动yarn，执行start-yarn.sh 命令 

[hadoop@cloud01 HDFSdemo]$ start-yarn.sh

[hadoop@cloud01 HDFSdemo]$ jps

22510  NodeManager

3、分布式执行wc.jar 

[hadoop@cloud01 ~]$ hadoop jar workspace/HDFSdemo/wc.jar mapreduce.WordCount 

3.1 执行过程日志情况

-- 连接ResourceManager： client.RMProxy: Connecting to ResourceManager

-- 获取分片，每个分片对应一个Map任务：input.FileInputFormat: Total input paths to process : 1

--生成本次运行的job编码：mapreduce.JobSubmitter: Submitting tokens for job: job_1424843731958_0002

--运行要执行的jar文件：mapreduce.Job: Running job: job_1424843731958_0002

--显示map和reduce执行进度

15/02/24 22:09:30 INFO mapreduce.Job:  map 0% reduce 0%
15/02/24 22:09:39 INFO mapreduce.Job:  map 100% reduce 0%
15/02/24 22:09:52 INFO mapreduce.Job:  map 100% reduce 100%
15/02/24 22:09:53 INFO mapreduce.Job: Job job_1424843731958_0002 completed successfully

3.2 MR整个过程的进程变化情况

ResourceManage，NodeManager->RunJar->MRAppMaster->YarnChild

随着MR程序进度的执行，响应的进程也随着退出，退出的顺序为

YarnChild->MRAppMaster->RunJar

3.3 图形方式给出对应的处理流程

图1

图2

file:/tmp/hadoop-hadoop/mapred/staging/hadoop1721666591/.staging/job_local1721666591_0001

常见问题

1、INFO ipc.Client: Retrying connect to server: cloud01/192.168.2.31:8032. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)

该问题是因为yarn没有启动，需要执行start-yarn.sh