[
https://issues.apache.org/jira/browse/HAMA-374?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13016731#comment-13016731
]
Thomas Jungblut commented on HAMA-374:
--------------------------------------
Not at all.
Syncing wasn't a great problem, I used a CountDownLatch for that. Works pretty
good :)
The problem I have now, is that every BSP thinks that he is the master groom.
This is actually not wrong because there's just one groom instance for all
runners.
So they always run through the {noformat}if
(bspPeer.getPeerName().equals(masterTask)) {noformat} passage.
I guess I rewrite it and start multiple local grooms and simulate a cluster and
let them do the multithreading. I guess this was what you mean with
multi-threaded.
> Add LocalBSPRunner for local testing of BSPs
> --------------------------------------------
>
> Key: HAMA-374
> URL: https://issues.apache.org/jira/browse/HAMA-374
> Project: Hama
> Issue Type: Improvement
> Components: bsp
> Affects Versions: 0.2.0
> Reporter: Thomas Jungblut
> Assignee: Thomas Jungblut
> Priority: Minor
> Fix For: 0.3.0
>
> Attachments: hama-374.patch
>
>
> Wanted to share it with you.
> I already implemented this as a subclass of the JobSubmissionProtocol. Just
> setting the jobSubmitClient in the BSPJobClient to a local version if the
> master adress is null.
> It is a simple single threaded running utility.
> Have a look at it!
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
