[
https://issues.apache.org/jira/browse/CHUKWA-430?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12790483#action_12790483
]
Jiaqi Tan commented on CHUKWA-430:
----------------------------------
+1 to patch, looks good to me
> Narrow down input for FSM mapreduce job
> ---------------------------------------
>
> Key: CHUKWA-430
> URL: https://issues.apache.org/jira/browse/CHUKWA-430
> Project: Hadoop Chukwa
> Issue Type: Bug
> Components: MR Data Processors
> Affects Versions: 0.4.0
> Reporter: Eric Yang
> Assignee: Eric Yang
> Fix For: 0.4.0
>
> Attachments: CHUKWA-430.patch
>
>
> FSMDataloader supplies all demux output data to FSM state machine. This is
> not efficient because most of the data type do not contribute to state
> generation. According to Jiaqi, the state machine requires the following
> types:
> JobHistoryTaskDataMapper:
> /chukwa/repos/chukwa/JobData
> /chukwa/repos/chukwa/TaskData
> TaskTrackerClientTraceMapper:
> /chukwa/repos/chukwa/ClientTraceDetailed
> DataNodeClientTraceMapper:
> /chukwa/repos/chukwa/ClientTraceDetailed
> This jira is to optimize the data loader supplied input, and narrow down the
> required input type.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
