[
https://issues.apache.org/jira/browse/HBASE-18649?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16180269#comment-16180269
]
ramkrishna.s.vasudevan commented on HBASE-18649:
------------------------------------------------
bq.So, a MR job that is from branch-1 going against a branch-2 cluster will
fail? I think that violates our attempt at making it so a branch-1 client can
work against a branch-2 cluster?
Am still not sure what will be the client code looking like in this case. For
eg. if there is a custom MR job written that tries to read a set of Hfiles and
then dump them to another table. Ideally the custom MR would have used the
KeyValueMapper/Reducer and in the client code they would have set KeyValue as
the output type. Then it should work. No problem. (because we are still having
the KV based mapper and reducer in the code base).
But if a client (custom MR job) is using HBase's Import tool and the output of
the reducer it expects KeyValue then it will fail. But one arguement is that
KeyValue is actually @Private (but the counter arg is that these KVMappers and
Reducers) should not have been exposed with KeyValue then :).
Are you good with the above plan of including all the KV based mappers/reducers
in the branch-2 code with @deprecated tags over them?
> Deprecate KV Usage in MR to move to Cells in 3.0
> ------------------------------------------------
>
> Key: HBASE-18649
> URL: https://issues.apache.org/jira/browse/HBASE-18649
> Project: HBase
> Issue Type: Improvement
> Components: mapreduce
> Affects Versions: 2.0.0-alpha-2
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
> Fix For: 2.0.0, 3.0.0, 2.1.0, 2.0.0-alpha-4
>
> Attachments: HBASE-18649_branch-2.patch, HBASE-18649_master_2.patch,
> HBASE-18649_master_3.patch, HBASE-18649_master_5.patch,
> HBASE-18649_master_6.patch, HBASE-18649_master.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
