What is the amount of data you are attempting to crunch in one MR job? Note that Map intermediate outputs are written to disk before being sent to reducers and this counts for non-DFS usage. So to say grossly, if your input is 14 GB, you surely need more than 2 or 3 x 14G free space overall to do the whole process.


On Thu, Jan 17, 2013 at 7:20 PM, Vikas Jadhav <vikascjadhav87@gmail.com> wrote:
Here is my problem 
I am using bulk loading for Hbase using MapReduce Program
 
Configured Capacity : 15.5 GB
DFS Used : 781.91 MB
Non DFS Used : 1.68 GB
DFS Remaining : 13.06 GB
DFS Used% : 4.93 %
DFS Remaining% : 84.26 %
 
But when i run my program
 
Configured Capacity : 15.5 GB
DFS Used : 819.69 MB
Non DFS Used : 14.59 GB
DFS Remaining : 116.01 MB
DFS Used% : 5.16 %
DFS Remaining% : 0.73 %
 
I have disable WAL in hbase  still its consuming non-dfs 
and  my program fails have tried lot times but no luck
 
SO WHAT SHLOULD I DO SO THAT NON DFS WILL NOT CONSUME WHOLE SPACE
 
I AM ALSO NOT ABLE TO FIND REASON BEHIND usage of non-dfs space to this large extent
 
 
13/01/17 08:44:07 INFO mapred.JobClient:  map 83% reduce 22%
13/01/17 08:44:09 INFO mapred.JobClient:  map 84% reduce 22%
13/01/17 08:44:12 INFO mapred.JobClient:  map 85% reduce 22%
13/01/17 08:44:15 INFO mapred.JobClient:  map 86% reduce 22%
13/01/17 08:44:18 INFO mapred.JobClient:  map 87% reduce 22%
13/01/17 08:44:22 INFO mapred.JobClient:  map 79% reduce 22%
13/01/17 08:44:25 INFO mapred.JobClient:  map 80% reduce 25%
13/01/17 08:44:27 INFO mapred.JobClient: Task Id : attempt_201301170837_0004_m_000009_0, Status : FAILED
FSError: java.io.IOException: No space left on device
java.lang.Throwable: Child Error
        at org.apache.hadoop.mapred.TaskRunner.run(TaskRunner.java:271)
Caused by: java.io.IOException: Creation of /tmp/hadoop-cfgsas1/mapred/local/userlogs/job_201301170837_0004/attempt_201301170837_0004_m_000009_0.cleanup failed.
        at org.apache.hadoop.mapred.TaskLog.createTaskAttemptLogDir(TaskLog.java:104)
        at org.apache.hadoop.mapred.DefaultTaskController.createLogDir(DefaultTaskController.java:71)
        at org.apache.hadoop.mapred.TaskRunner.prepareLogFiles(TaskRunner.java:316)
        at org.apache.hadoop.mapred.TaskRunner.run(TaskRunner.java:228)
13/01/17 08:44:27 WARN mapred.JobClient: Error reading task outputhttp://rdcesx12078.race.sas.com:50060/tasklog?plaintext=true&attemptid=attempt_201301170837_0004_m_000009_0&filter=stdout
13/01/17 08:44:27 WARN mapred.JobClient: Error reading task outputhttp://rdcesx12078.race.sas.com:50060/tasklog?plaintext=true&attemptid=attempt_201301170837_0004_m_000009_0&filter=stderr
13/01/17 08:44:28 INFO mapred.JobClient:  map 82% reduce 25%
13/01/17 08:44:31 INFO mapred.JobClient:  map 83% reduce 25%
13/01/17 08:45:07 INFO mapred.JobClient:  map 83% reduce 27%
 



On Wed, Jan 16, 2013 at 6:43 PM, Jean-Marc Spaggiari <jean-marc@spaggiari.org> wrote:
I think you can still run with the OS on another drive, or on a live
USB drive, or even on the memory only, loaded from the network while
the server is booting from the network drive, etc. No?

JM

2013/1/16, Mohammad Tariq <dontariq@gmail.com>:
> That would be really cool Chris.
> +1 for that.
>
> Warm Regards,
> Tariq
> https://mtariq.jux.com/
> cloudfront.blogspot.com
>
>
> On Wed, Jan 16, 2013 at 6:15 PM, Chris Embree <cembree@gmail.com> wrote:
>
>> Ha, you joke, but we're planning on running with no local OS.  If it
>> works
>> as planned I'll post a nice summary of our approach. :)
>>
>>
>> On Wed, Jan 16, 2013 at 2:53 AM, Harsh J <harsh@cloudera.com> wrote:
>>
>>> <kidding> Wipe your OS out. </kidding>
>>>
>>> Please read: http://search-hadoop.com/m/9Qwi9UgMOe
>>>
>>>
>>> On Wed, Jan 16, 2013 at 1:16 PM, Vikas Jadhav
>>> <vikascjadhav87@gmail.com>wrote:
>>>
>>>>
>>>> how to remove non dfs space from  hadoop cluster
>>>>
>>>> --
>>>> *
>>>> *
>>>> *
>>>>
>>>> Thanx and Regards*
>>>> * Vikas Jadhav*
>>>>
>>>
>>>
>>>
>>> --
>>> Harsh J
>>>
>>
>>
>



--


Thanx and Regards
 Vikas Jadhav



--
Harsh J