如何使用Hadoop Streaming在本地Hadoop集群中运行MRJob? [英] How to run a MRJob in a local Hadoop Cluster with Hadoop Streaming?

问题描述

我目前正在上大数据课，我的一个项目是在本地设置的Hadoop集群上运行我的Mapper/Reducer.

I'm currently taking a Big Data Class, and one of my projects is to run my Mapper/Reducer on a Hadoop Cluster which is set up locally.

我一直在使用Python和MRJob库作为类.

I've been using Python along with the MRJob library for the class.

这是我当前用于Mapper/Reducer的Python代码.

Here is my current Python Code for the Mapper/Reducer.

from mrjob.job import MRJob
from mrjob.step import MRStep
import re
import os

WORD_RE = re.compile(r"[\w']+")
choice = ""

class MRPrepositionsFinder(MRJob):

def steps(self):
    return [
        MRStep(mapper=self.mapper_get_words),
        MRStep(reducer=self.reducer_find_prep_word)
    ]

def mapper_get_words(self, _, line):
    # set word_list to indicators, convert to lowercase, and strip whitespace
    word_list = set(line.lower().strip() for line in open("/hdfs/user/user/indicators.txt"))

    # set filename to map_input_file
    fileName = os.environ['map_input_file']
    # itterate through each word in line
    for word in WORD_RE.findall(line):
        # if word is in indicators, yield chocie as filename
        if word.lower() in word_list:
            choice = fileName.split('/')[5]
            yield (choice, 1)

def reducer_find_prep_word(self, choice, counts):
    # each item of choice is (choice, count),
    # so yielding results in value=choice, key=count
    yield (choice, sum(counts))


if __name__ == '__main__':
MRPrepositionsFinder.run()

当我尝试在Hadoop集群上运行代码时-我使用了以下命令:

When I try to run the code on my Hadoop Cluster - I used the following command:

python hrc_discover.py /hdfs/user/user/HRCmail/* -r hadoop --hadoop-bin /usr/bin/hadoop > /hdfs/user/user/output

不幸的是，每次我运行命令时，都会出现以下错误:

Unfortunately every time I run the command I get the following error:

No configs found; falling back on auto-configuration
STDERR: Error: JAVA_HOME is not set and could not be found.
Traceback (most recent call last):
  File "hrc_discover.py", line 37, in 
    MRPrepositionsFinder.run()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/job.py", line 432, in run
    mr_job.execute()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/job.py", line 453, in execute
    super(MRJob, self).execute()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/launch.py", line 161, in execute
    self.run_job()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/launch.py", line 231, in run_job
    runner.run()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/runner.py", line 437, in run
    self._run()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/hadoop.py", line 346, in _run
    self._find_binaries_and_jars()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/hadoop.py", line 361, in _find_binaries_and_jars
    self.get_hadoop_version()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/hadoop.py", line 198, in get_hadoop_version
    return self.fs.get_hadoop_version()
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/fs/hadoop.py", line 117, in get_hadoop_version
    stdout = self.invoke_hadoop(['version'], return_stdout=True)
  File "/usr/lib/python3.5/site-packages/mrjob-0.6.0.dev0-py3.5.egg/mrjob/fs/hadoop.py", line 172, in invoke_hadoop
    raise CalledProcessError(proc.returncode, args)
subprocess.CalledProcessError: Command '['/usr/bin/hadoop', 'version']' returned non-zero exit status 1

我环顾了互联网，发现我需要导出JAVA_HOME变量-但我不想设置任何可能破坏设置的内容.

I looked around the internet and found out that I need to export my JAVA_HOME variable - but I don't want to set anything that might break my setup.

任何帮助，将不胜感激，谢谢！

Any help with this would be much appreciated, thanks!

推荐答案

问题似乎出在 etc/hadoop/hadoop-env.sh 脚本文件中.

It seems like the issue was in the etc/hadoop/hadoop-env.sh script file.

JAVA_HOME 环境变量配置为:

export JAVA_HOME=$(JAVA_HOME)

所以我继续将其更改为以下内容:

So I went ahead and changed it to the following:

export JAVA_HOME=/usr/lib/jvm/java-8-openjdk

我试图再次运行以下命令，希望它能起作用:

I attempted to run the following command again, in hopes that it would work:

python hrc_discover.py /hdfs/user/user/HRCmail/* -r hadoop --hadoop-bin /usr/bin/hadoop > /hdfs/user/user/output

非常感谢MRJob在JAVA_HOME环境中的学习，并产生了以下输出:

Thankfully MRJob picked up on the JAVA_HOME environment and resulted in the following output:

No configs found; falling back on auto-configuration
Using Hadoop version 2.7.3
Looking for Hadoop streaming jar in /home/hadoop/contrib...
Looking for Hadoop streaming jar in /usr/lib/hadoop-mapreduce...
Hadoop streaming jar not found. Use --hadoop-streaming-jar
Creating temp directory /tmp/hrc_discover.user.20170306.022649.449218
Copying local files to hdfs:///user/user/tmp/mrjob/hrc_discover.user.20170306.022649.449218/files/...
.. 

要解决Hadoop流媒体jar的问题，我在命令中添加了以下开关:

To fix the issue with the Hadoop streaming jar, I added the following switch to the command:

--hadoop-streaming-jar /usr/lib/hadoop/share/hadoop/tools/lib/hadoop-streaming-2.7.3.jar

完整的命令如下所示:

python hrc_discover.py /hdfs/user/user/HRCmail/* -r hadoop --hadoop-streaming-jar /usr/lib/hadoop/share/hadoop/tools/lib/hadoop-streaming-2.7.3.jar --hadoop-bin /usr/bin/hadoop > /hdfs/user/user/output

结果如下:

No configs found; falling back on auto-configuration
Using Hadoop version 2.7.3
Creating temp directory /tmp/hrc_discover.user.20170306.022649.449218
Copying local files to hdfs:///user/user/tmp/mrjob/hrc_discover.user.20170306.022649.449218/files/...

问题似乎已经解决，Hadoop应该处理我的工作.

It seems the issue has been resolved and Hadoop should process my job.

这篇关于如何使用Hadoop Streaming在本地Hadoop集群中运行MRJob?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！