[
https://issues.apache.org/jira/browse/HAMA-557?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13419341#comment-13419341
]
Thomas Jungblut commented on HAMA-557:
--------------------------------------
Took a bit longer because I had to checkout the branch. Sorry for that.
Questions:
-Compile Problem TestCheckpoint overrides the method replayMessages, but it
does not exist, should it exist?
-In hama-core there is a folder created called "nullzookeeper", from what
testcase does that come from?
-Can we remove the tilde's ~ from debug output?
-Do you think we should stick with defining interfaces with "I" in front? I'm
naming interfaces without them and call the concrete implementations *Impl.
What do you think is the best?
-Now we have a lot of services, we could extract init and close to a
superinterface, WDYT? Don't know about the usage then if they can be composed.
Additional notes:
-A few imports could be organized
-Leave out the abstract declarations in interfaces (MasterSyncClient for
example)
-In JobInProgress you are using newInstance several times with new Object[0], I
think this is not needed allocation, the method is null save as far as I know,
so you can simply pass NULL
-Several // TODO Auto-generated catch block comments
(this is actually the most important note)
-in checkpoint received messages, you are writing the classname with each
message, I think we should combine messages by class and then write it.
Something like the message bundles does.
Testcases are fine. Thanks for the great work Suraj :)
> Implement Checkpointing service in Hama
> ---------------------------------------
>
> Key: HAMA-557
> URL: https://issues.apache.org/jira/browse/HAMA-557
> Project: Hama
> Issue Type: Sub-task
> Components: bsp core
> Affects Versions: 0.6.0
> Reporter: Suraj Menon
> Assignee: Suraj Menon
> Fix For: 0.6.0
>
> Attachments: HAMA-505-557-610-611-v1.patch,
> HAMA-557-ft-framework.patch
>
>
> Implement checkpointing service in Apache Hama. My patches for HAMA-533 and
> HAMA-534 are blocked on this.
> - Checkpointing should be done as messages are either sent or received. I
> prefer while receiving messages, as we can achieve some parallelism with
> asynchronous messages. Please comment if you differ.
> - BSPMaster should hold the checkpoint status for each task. Checkpoint
> status includes superstep count and file information for which checkpointing
> is complete
> - MessageManager should notify Checkpointer of a new message at BSPPeer.
> - Implement/Reuse MessageBundle class as splitClass in BSPPeerImpl for
> recovery in initInput.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
