incubator-hama-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Edward J. Yoon" <edwardy...@apache.org>
Subject Re: Share TODO-list for hama-0.2
Date Thu, 10 Feb 2011 08:33:19 GMT
Because, the unit tests of TestBSPMaster are almost commented text at
this time. We have to uncomment and fix bugs.

P.S. Now I'm starting the implement of kill job method and I have a
question about your code.

To kill the child processes, GroomServer have to know which tasks are
no longer valid and should be killed. As you know we removed some data
structures e.g., taskIdToGroomNameMap, TaskIdToTaskInProgressMap, ...
from BSPMaster.

So, I'm thinking, 1) checks the job's state while GroomServer running,
2) and kills and cleans them.

Then, same as it was based on heart-beat communication between
BSPMaster and GroomServer. You got a better idea?

On Thu, Feb 10, 2011 at 4:39 PM, Chia-Hung Lin <clin4j@googlemail.com> wrote:
> 2011/2/9 Edward J. Yoon <edwardyoon@apache.org>:
>> Hi all,
>>
>> Here is our TODO-list:
>>
>>  1) Below job commands are not implemented yet
>>
>>        [-kill <job-id>]
>>        [-list-attempt-ids <job-id> <task-state>]
>>        [-kill-task <task-id>]
>>        [-fail-task <task-id>]
>>
>> 2) TestBSPMaster doesn't work.
>>
>> It seems like a bug by adding umbilical interface to GroomServer.
>> Anyone can figure out this problem?
>
> Is there any log information available? The unit test on my machine
> looks working as it reports:
>
> test:
>    ...
>    [junit] Running org.apache.hama.bsp.TestBSPMaster
>    [junit] Tests run: 1, Failures: 0, Errors: 0, Time elapsed: 6.248 sec
>    [junit] Running org.apache.hama.bsp.TestBSPPeer
>    [junit] Tests run: 1, Failures: 0, Errors: 0, Time elapsed: 71.882 sec
>    ...
> BUILD SUCCESSFUL
> Total time: 1 minute 51 seconds
>
> But when testing with SerializePrinting example, it would returns
>
>    java.io.FileNotFoundException: File does not exist: /tmp/test-example/0
>
> exception. Seems that SerializePrinting will delete test-example dir
> even if it exists.
>
>>
>> 3) Lack of comments and documentations
>>
>> 4) Increase committers
>>
>> --
>> Best Regards, Edward J. Yoon
>> http://blog.udanax.org
>> http://twitter.com/eddieyoon
>>
>



-- 
Best Regards, Edward J. Yoon
http://blog.udanax.org
http://twitter.com/eddieyoon

Mime
View raw message