[
https://issues.apache.org/jira/browse/HBASE-11572?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14071193#comment-14071193
]
Nick Dimiduk commented on HBASE-11572:
--------------------------------------
Isn't it better to leave the client ignorant of specific replicas? It makes
sense for tests, but how is it helpful for prod usecases? Maybe you have an
example?
> Add support for doing get/scans against a particular replica_id
> ---------------------------------------------------------------
>
> Key: HBASE-11572
> URL: https://issues.apache.org/jira/browse/HBASE-11572
> Project: HBase
> Issue Type: Sub-task
> Reporter: Enis Soztutar
>
> We have support for doing an rpc against the primary region in get/scans and
> also new functionality for doing failover get/scan RPCs trying primary first
> then doing backup rpcs to secondaries when the first one times out.
> We can add functionality for doing get/scan against a particular replica_id
> so that we can use it in tests, or by clients in some specific situations.
> We can think about adding support for querying the replica_id from get/scan
> responses coming from failover rpcs so that the client can then issue a
> follow up request against that particular replica_id.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
