hadoop-common-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefan Groschupf ...@101tec.com>
Subject Re: [memory leak?] Re: MapReduce failure
Date Sun, 16 Mar 2008 07:15:44 GMT
I do not instantiate 3 GB of objects, that is for sure. The wordcount  
example does not run anymore so I dont think this is something  
releated to my code, beside the wordcount example many other users  
report the same problem:
See:
http://markmail.org/search/?q=org.apache.hadoop.mapred.MapTask%24MapOutputBuffer.collect+order%3Adate-backward
Thanks for your help!

Stefan


On Mar 15, 2008, at 11:02 PM, Devaraj Das wrote:

> It might have something to do with your application itself. By any  
> chance
> are you doing a lot of huge object allocation (directly or indirectly)
> within the map method? Which version of hadoop are you on?
>
>> -----Original Message-----
>> From: Stefan Groschupf [mailto:sg@101tec.com]
>> Sent: Sunday, March 16, 2008 9:07 AM
>> To: core-user@hadoop.apache.org
>> Subject: [memory leak?] Re: MapReduce failure
>>
>> Hi there,
>>
>> we see the same situation and browsing the posts there are
>> quite a lot of people running into this OOM problem.
>> We run a own Mapper and our mapred.child.java.opts is
>> -Xmx3048m, I think that should be more then enough.
>> Also I changed io.sort.mb to 10, which had also no impact.
>>
>> Any ideas what might cause the OutOfMemoryError ?
>> Thanks.
>> Stefan
>>
>>
>>
>>
>> On Mar 9, 2008, at 10:28 PM, Amar Kamat wrote:
>>
>>> What is the heap size you are using for your tasks? Check
>>> 'mapred.child.java.opts' in your hadoop-default.xml. Try increasing
>>> it. This will happen if you try running the random-writer + sort
>>> examples with default parameters. The maps are not able to
>> spill the
>>> data to the disk. Btw what version of HADOOP are you using?
>>> Amar
>>> On Mon, 10 Mar 2008, Ved Prakash wrote:
>>>
>>>> Hi friends,
>>>>
>>>> I have made a cluster of 3 machines, one of them is
>> master, and other
>>>> 2 slaves. I executed a mapreduce job on master but after Map, the
>>>> execution terminates and Reduce doesn't happen. I have checked dfs
>>>> and no output folder gets created.
>>>>
>>>> this is the error I see
>>>>
>>>> 08/03/10 10:35:21 INFO mapred.JobClient: Task Id :
>>>> task_200803101001_0001_m_000064_0, Status : FAILED
>>>> java.lang.OutOfMemoryError: Java heap space
>>>>      at
>>>> java.io.ByteArrayOutputStream.write(ByteArrayOutputStream.java
>>>> :95)
>>>>      at java.io.DataOutputStream.write(DataOutputStream.java:90)
>>>>      at org.apache.hadoop.io.Text.write(Text.java:243)
>>>>      at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.collect(
>>>> MapTask.java:347)
>>>>      at org.apache.hadoop.examples.WordCount
>>>> $MapClass.map(WordCount.java
>>>> :72)
>>>>      at org.apache.hadoop.examples.WordCount
>>>> $MapClass.map(WordCount.java
>>>> :59)
>>>>      at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:50)
>>>>      at org.apache.hadoop.mapred.MapTask.run(MapTask.java:192)
>>>>      at org.apache.hadoop.mapred.TaskTracker
>>>> $Child.main(TaskTracker.java
>>>> :1787)
>>>>
>>>> 08/03/10 10:35:22 INFO mapred.JobClient:  map 55% reduce
>> 17% 08/03/10
>>>> 10:35:31 INFO mapred.JobClient:  map 56% reduce 17%
>> 08/03/10 10:35:51
>>>> INFO mapred.JobClient:  map 57% reduce 17% 08/03/10 10:36:04 INFO
>>>> mapred.JobClient:  map 58% reduce 17% 08/03/10 10:36:07 INFO
>>>> mapred.JobClient:  map 57% reduce 17% 08/03/10 10:36:07 INFO
>>>> mapred.JobClient: Task Id :
>>>> task_200803101001_0001_m_000071_0, Status : FAILED
>>>> java.lang.OutOfMemoryError: Java heap space
>>>>      at
>>>> java.io.ByteArrayOutputStream.write(ByteArrayOutputStream.java
>>>> :95)
>>>>      at java.io.DataOutputStream.write(DataOutputStream.java:90)
>>>>      at org.apache.hadoop.io.Text.write(Text.java:243)
>>>>      at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.collect(
>>>> MapTask.java:347)
>>>>      at org.apache.hadoop.examples.WordCount
>>>> $MapClass.map(WordCount.java
>>>> :72)
>>>>      at org.apache.hadoop.examples.WordCount
>>>> $MapClass.map(WordCount.java
>>>> :59)
>>>>      at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:50)
>>>>      at org.apache.hadoop.mapred.MapTask.run(MapTask.java:192)
>>>>      at org.apache.hadoop.mapred.TaskTracker
>>>> $Child.main(TaskTracker.java
>>>> :1787)
>>>>
>>>> though it tries to overcome this problem but the mapreduce
>>>> application doesn't create output, can anyone tell me why is this
>>>> happening?
>>>>
>>>> Thanks
>>>>
>>>
>>
>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> 101tec Inc.
>> Menlo Park, California, USA
>> http://www.101tec.com
>>
>>
>>
>
>

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
101tec Inc.
Menlo Park, California, USA
http://www.101tec.com



Mime
View raw message