[
https://issues.apache.org/jira/browse/MAPREDUCE-3966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14314987#comment-14314987
]
Craig Welch commented on MAPREDUCE-3966:
----------------------------------------
Hmm, I actually thing we would be better off pulling out all of this complex
storage / caching logic and defer that to a general purpose database layer, as
suggested here:
https://issues.apache.org/jira/browse/MAPREDUCE-3973?focusedCommentId=13226450&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13226450
> Add separate cache for CompletedJobs with tasks loaded and without tasks
> loaded in JobHIstoryServer
> ---------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3966
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3966
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: jobhistoryserver, mrv2
> Affects Versions: 0.23.1
> Reporter: Vinod Kumar Vavilapalli
>
> After MAPREDUCE-3901, we should have separate caches for list of CompletedJob
> with tasks loaded and list of CompletedJob without tasks loaded. Once we have
> separate cache, we should increase the cache size for the list which doesn't
> load tasks.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
