hadoop流作业在python中失败

Question

im trying to implement an algorithm in hadoop. i tried to execute part of the code in hadoop but streaming job fails

$ /home/hadoop/hadoop/bin/hadoop jar contrib/streaming/hadoop-*-streaming.jar -file /home/hadoop/hadoop/PR/mapper.py -mapper mapper.py -file /home/hadoop/hadoop/PR/reducer.py -reducer reducer.py -input pagerank/* -output PRoutput6

packageJobJar: [/home/hadoop/hadoop/PR/mapper.py, /home/hadoop/hadoop/PR/reducer.py, /home/hadoop/hadoop/tmp/dir/hadoop-hadoop/hadoop-unjar7101759175212283428/] [] /tmp/streamjob6286075675343269479.jar tmpDir=null

11/04/23 01:03:24 INFO mapred.FileInputFormat: Total input paths to process : 1

11/04/23 01:03:24 INFO streaming.StreamJob: getLocalDirs(): [/home/hadoop/hadoop/tmp/dir/hadoop-hadoop/mapred/local]

11/04/23 01:03:24 INFO streaming.StreamJob: Running job: job_201104222325_0021

11/04/23 01:03:24 INFO streaming.StreamJob: To kill this job, run:

11/04/23 01:03:24 INFO streaming.StreamJob: /home/hadoop/hadoop/bin/../bin/hadoop job  -Dmapred.job.tracker=localhost:54311 -kill job_201104222325_0021

11/04/23 01:03:24 INFO streaming.StreamJob: Tracking URL: http://localhost:50030/jobdetails.jsp?jobid=job_201104222325_0021

11/04/23 01:03:25 INFO streaming.StreamJob:  map 0%  reduce 0%

11/04/23 01:03:31 INFO streaming.StreamJob:  map 50%  reduce 0%

11/04/23 01:03:41 INFO streaming.StreamJob:  map 50%  reduce 17%

11/04/23 01:03:56 INFO streaming.StreamJob:  map 100%  reduce 100%

11/04/23 01:03:56 INFO streaming.StreamJob: To kill this job, run:

11/04/23 01:03:56 INFO streaming.StreamJob: /home/hadoop/hadoop/bin/../bin/hadoop job  -Dmapred.job.tracker=localhost:54311 -kill job_201104222325_0021

11/04/23 01:03:56 INFO streaming.StreamJob: Tracking URL: http://localhost:50030/jobdetails.jsp?jobid=job_201104222325_0021

11/04/23 01:03:56 ERROR streaming.StreamJob: Job not Successful!

11/04/23 01:03:56 INFO streaming.StreamJob: killJob...

Streaming Job Failed!

mapper.py

#!/usr/bin/env python
import sys
import itertools

def ipsum(input_key,input_value_list):
   return sum(input_value_list)

n= 20 # works up to about 1000000 pages
i = {}
for j in xrange(n): i[j] = [1.0/n,0,[]]
j=0
u=0
for line in sys.stdin:
  if j<n:
    i[j][1]=int(line)
  j=j+1

  if j > n: 
    if line != "-1\n":
      i[u][2] = line.split(',')
    else: 
      i[u][2]=[]
    u=u+1
for j in xrange(n):
  if i[j][1] != 0:
    i[j][2] = map(int,i[j][2])    

intermediate=[]
for (input_key,input_value) in i.items():
  if input_value[1] == 0: intermediate.extend([(1,input_value[0])])
  else: intermediate.extend([])
grp = {}
for key, group in itertools.groupby(sorted(intermediate),lambda x: x[0]):
  grp[key] = list([y for x, y in group])
iplist = [ipsum(intermediate_key,grp[intermediate_key]) for intermediate_key in grp]
inter=[]
for (input_key,input_value) in i.items():
  if input_value[1] == 0: inter.extend([(input_key,0.0)]+[(outlink,input_value[0]/input_value[1]) for outlink in input_value[2]])
  else: inter.extend([])

for value in inter:
  value1 = value[0]
  value2 = value[1]
  print '%s %s' % (value1,value2)

reducer.py

#!/usr/bin/env python
import sys
import itertools
for line in sys.stdin:
  input_key, input_value=line.split(' ',1)
  input_key = input_key.strip()
  input_value = input_value.strip()
  input_key = int(input_key)
  input_value = float(input_value)
  print str(input_key)+' '+str(input_value)

i dont know whether the error is in my code or hadoop config... because i was able to execute the code using, $ cat /home/hadoop/hadoop/PR/pagerank/input.txt | python /home/hadoop/hadoop/PR/mapper.py | sort | python /home/hadoop/hadoop/PR/reducer.py

would appreciate any help, Thank you.

Answer 1

Take a look at the job info page url from the output. In your case, localhost:50030/jobdetails.jsp?jobid=job_201104222325_0021

Click on the number in the "failed mappers" column and the "last 8KB" (or whatever) log link and you will see the (most likely) python exception you're hitting.

Answer 2

My guess is your data may be the key. Casting a float from a string or similar issue may be hitting a bumb in your real data that does not appear in your local test data. Perhaps you could address with exception handling or assertions.