[
https://issues.apache.org/jira/browse/HBASE-18448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16121061#comment-16121061
]
Anoop Sam John commented on HBASE-18448:
----------------------------------------
Ya the 2nd approach. I just want to correct one of ur understanding. One EP
execution from client side can NOT get executed in all region irrespective of
the table. The CPEP call itself has to happen on a table. You can specify a row
range though. So what happens is the client side will find all the regions of
this table coming under the specified row range and do RPC to all such regions.
As u said the execution across regions will happen in parallel way. You can
pass the start and end keys as empty byte[] (HConstants.EMPTY_START_ROW) when
we need the execution to happen across ALL the regions of the table. So there
is no need to check on a region whether this belongs to specified table or not.
In fact passing the table name to server is not needed. When the system do
execute the CPEP for a region it is sure that this region belongs to the table
that u have given. Hope u are fully clear now :-)
Ya if u can expose the CPEP client side API for ur user, there is no need for
HBase side changes. On the Public Table interface better we do NOT expose such
very special APIs. In ur CPEP impl (server side), u will get handle to current
Region and from that u can get Stores and on Store interface u can call the
refresh method.
> Added support for refreshing HFiles through API and shell
> ---------------------------------------------------------
>
> Key: HBASE-18448
> URL: https://issues.apache.org/jira/browse/HBASE-18448
> Project: HBase
> Issue Type: Sub-task
> Affects Versions: 2.0.0, 1.3.1
> Reporter: Ajay Jadhav
> Assignee: Ajay Jadhav
> Priority: Minor
> Fix For: 1.4.0
>
> Attachments: HBASE-18448.branch-1.001.patch,
> HBASE-18448.branch-1.002.patch
>
>
> In the case where multiple HBase clusters are sharing a common rootDir, even
> after flushing the data from
> one cluster doesn't mean that other clusters (replicas) will automatically
> pick the new HFile. Through this patch,
> we are exposing the refresh HFiles API which when issued from a replica will
> update the in-memory file handle list
> with the newly added file.
> This allows replicas to be consistent with the data written through the
> primary cluster.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
