[
https://issues.apache.org/jira/browse/MAPREDUCE-1376?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12803585#action_12803585
]
Hong Tang commented on MAPREDUCE-1376:
--------------------------------------
bq. I agree that the strategy for obtaining the list of users is a reasonable
abstraction, but why would this be part of the driver and not the user
resolution types?
My thoughts is that the knowledge of the list of targeted users is tied to a
specific cluster where we run gridmix and thus should be independent from which
specific resolver we use. But on the other hand, the interface I suggested is
also less flexible that it prevents us from passing down more info to the
resolver in its initialization (e.g. what if we want to do weighted random
among the set of targeted users). So I am leaning toward your suggestion of
keeping the URI-based interface but making a note that commonly, the URI should
point to a file following the simple format parsable by
UserResolver.parseUserList (would you please add such comments in the javadoc
for UserResolver.setTargetUsers?).
bq. It seems most of our disagreement would go away if I just removed the
-users parameter and all resolution was done through Configuration.
I actually do not like the idea of completely hiding the abstraction of
targeted users. Would you agree?
So in summary, I am fine with the latest patch with only the minor suggestion
of adding a comment to emphasize the common format convention of userspec.
> Support for varied user submission in Gridmix
> ---------------------------------------------
>
> Key: MAPREDUCE-1376
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-1376
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: contrib/gridmix
> Reporter: Chris Douglas
> Assignee: Chris Douglas
> Attachments: M1376-0.patch, M1376-1.patch, M1376-2.patch,
> M1376-3.patch
>
>
> Gridmix currently submits all synthetic jobs as the client user. It should be
> possible to map users in the trace to a set of users appropriate for the
> target cluster.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
