You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hama.apache.org by "Edward J. Yoon" <ed...@apache.org> on 2010/10/14 07:12:16 UTC
[DISCUSS] 0.2.0 feature freeze.
Hello all,
Recently, TRUNK has progressed quite a bit. It consist of ..
- BSP computing framework and its programming model.
- Simple command line Job management tool
- Example, a BSP based Pi estimation
I'd like to propose a feature freeze for hama-0.2.0 release.
Of course, there are several (at least) things we might want to do
before releasing. For examples, 1) current "GroomServer" manages only
one instance of BSPPeer, we have to improve it to be able to manage
multiple instances of BSPPeer. 2) There is still no FT (fault
tolerant) mechanism for slave node failure. etc. But, the problem is
that it would take too long time.
What do you think?
--
Best Regards, Edward J. Yoon
edwardyoon@apache.org
http://blog.udanax.org
Re: [DISCUSS] 0.2.0 feature freeze.
Posted by "Edward J. Yoon" <ed...@apache.org>.
Then now, TODO things are as below:
- implement job kill() method
- add unit tests
- implement superstep counter
- implement TaskRunner
- documentation and update website
And, un-known bugs.
Thanks.
On Fri, Oct 15, 2010 at 10:08 AM, Edward J. Yoon <ed...@apache.org> wrote:
>> 1) means multiple task slots for each groom server?
>
> Yup
>
> On Thu, Oct 14, 2010 at 9:29 PM, Filipe David Manana
> <fd...@apache.org> wrote:
>> On Thu, Oct 14, 2010 at 6:12 AM, Edward J. Yoon <ed...@apache.org> wrote:
>>> Of course, there are several (at least) things we might want to do
>>> before releasing. For examples, 1) current "GroomServer" manages only
>>> one instance of BSPPeer, we have to improve it to be able to manage
>>> multiple instances of BSPPeer. 2) There is still no FT (fault
>>> tolerant) mechanism for slave node failure. etc. But, the problem is
>>> that it would take too long time.
>>
>> 1) means multiple task slots for each groom server?
>>
>>>
>>> What do you think?
>>
>> +1
>>>
>>> --
>>> Best Regards, Edward J. Yoon
>>> edwardyoon@apache.org
>>> http://blog.udanax.org
>>>
>>
>>
>>
>> --
>> Filipe David Manana,
>> fdmanana@gmail.com, fdmanana@apache.org
>>
>> "Reasonable men adapt themselves to the world.
>> Unreasonable men adapt the world to themselves.
>> That's why all progress depends on unreasonable men."
>>
>
>
>
> --
> Best Regards, Edward J. Yoon
> edwardyoon@apache.org
> http://blog.udanax.org
>
--
Best Regards, Edward J. Yoon
edwardyoon@apache.org
http://blog.udanax.org
Re: [DISCUSS] 0.2.0 feature freeze.
Posted by "Edward J. Yoon" <ed...@apache.org>.
> 1) means multiple task slots for each groom server?
Yup
On Thu, Oct 14, 2010 at 9:29 PM, Filipe David Manana
<fd...@apache.org> wrote:
> On Thu, Oct 14, 2010 at 6:12 AM, Edward J. Yoon <ed...@apache.org> wrote:
>> Of course, there are several (at least) things we might want to do
>> before releasing. For examples, 1) current "GroomServer" manages only
>> one instance of BSPPeer, we have to improve it to be able to manage
>> multiple instances of BSPPeer. 2) There is still no FT (fault
>> tolerant) mechanism for slave node failure. etc. But, the problem is
>> that it would take too long time.
>
> 1) means multiple task slots for each groom server?
>
>>
>> What do you think?
>
> +1
>>
>> --
>> Best Regards, Edward J. Yoon
>> edwardyoon@apache.org
>> http://blog.udanax.org
>>
>
>
>
> --
> Filipe David Manana,
> fdmanana@gmail.com, fdmanana@apache.org
>
> "Reasonable men adapt themselves to the world.
> Unreasonable men adapt the world to themselves.
> That's why all progress depends on unreasonable men."
>
--
Best Regards, Edward J. Yoon
edwardyoon@apache.org
http://blog.udanax.org
Re: [DISCUSS] 0.2.0 feature freeze.
Posted by Filipe David Manana <fd...@apache.org>.
On Thu, Oct 14, 2010 at 6:12 AM, Edward J. Yoon <ed...@apache.org> wrote:
> Of course, there are several (at least) things we might want to do
> before releasing. For examples, 1) current "GroomServer" manages only
> one instance of BSPPeer, we have to improve it to be able to manage
> multiple instances of BSPPeer. 2) There is still no FT (fault
> tolerant) mechanism for slave node failure. etc. But, the problem is
> that it would take too long time.
1) means multiple task slots for each groom server?
>
> What do you think?
+1
>
> --
> Best Regards, Edward J. Yoon
> edwardyoon@apache.org
> http://blog.udanax.org
>
--
Filipe David Manana,
fdmanana@gmail.com, fdmanana@apache.org
"Reasonable men adapt themselves to the world.
Unreasonable men adapt the world to themselves.
That's why all progress depends on unreasonable men."
Re: [DISCUSS] 0.2.0 feature freeze.
Posted by Sangwon Seo <sm...@gmail.com>.
Hi,
Nice to hear than.
I agree that it's the time to freeze a feature for hama.
2010/10/14 Edward J. Yoon <ed...@apache.org>
> Hello all,
>
> Recently, TRUNK has progressed quite a bit. It consist of ..
>
> - BSP computing framework and its programming model.
> - Simple command line Job management tool
> - Example, a BSP based Pi estimation
>
> I'd like to propose a feature freeze for hama-0.2.0 release.
>
> Of course, there are several (at least) things we might want to do
> before releasing. For examples, 1) current "GroomServer" manages only
> one instance of BSPPeer, we have to improve it to be able to manage
> multiple instances of BSPPeer. 2) There is still no FT (fault
> tolerant) mechanism for slave node failure. etc. But, the problem is
> that it would take too long time.
>
> What do you think?
>
> --
> Best Regards, Edward J. Yoon
> edwardyoon@apache.org
> http://blog.udanax.org
>
--
Best Regards,
Sangwon Seo.
--------------------------------------------------------
Sangwon Seo, Ph.D Candidate
Computer Architecture Lab,
KAIST (Korea Advanced Institute of Science and Technology)
http://smiler-note.blogspot.com