hadoop-common-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Amar Kamat <ama...@yahoo-inc.com>
Subject Re: After hadoop jobtracker runs for a long time, jobtracker may comsume 100% cpu (one cpu core is fully consume), why?
Date Mon, 09 Feb 2009 06:10:24 GMT
Ruyue Ma wrote:
> Our hadoop version is checked out from truck version at 2008.10.22.
>
> When this problem appeared, our cluster didn't have any job running. The
> cluster was idle!
>   
How many jobs (total) were submitted to the cluster? What is the average 
size of each job? Have you changed any default parameters and if yes 
which ones?
There is an issue open on this (see 
https://issues.apache.org/jira/browse/HADOOP-4766) for more details). Is 
there any way to reproduce this?
Amar
>
> On Mon, Feb 9, 2009 at 1:56 PM, Ruyue Ma <mry.maillist@gmail.com> wrote:
>
>   
>> Hi, Everyone
>>
>> After hadoop jobtracker runs for a long time, jobtracker may comsume 100%
>> cpu (one cpu core is fully consume).
>>
>> We have encountered this problem sever times. When it happens, it is hardly
>> to submit job to jobtracker and visit the jobtracker web page.
>>
>> Anybody encounter this?
>>
>> is there an existed patch to  fix it?
>>
>> Thanks
>> Ruyue Ma
>> Baidu.com
>>
>>     
>
>   


Mime
View raw message