hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Lars Hofhansl (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-10309) Add support to delete empty regions in 0.94.x series
Date Thu, 06 Feb 2014 19:10:13 GMT

    [ https://issues.apache.org/jira/browse/HBASE-10309?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13893670#comment-13893670
] 

Lars Hofhansl commented on HBASE-10309:
---------------------------------------

Can't we reuse the hbck code in a more targeted way for a specific known region that is known
ahead of time to be empty (and it won't be written to or split, etc)?

I fear that this will come up over and over again with folks that store event data that is
access in a most-recent-only way and eventually will be purged; and this is otherwise a compelling
usecase for HBase.

There are other ways around this (salting, prefix hashing, bounded time windows, etc) but
all of those have other drawbacks.


> Add support to delete empty regions in 0.94.x series
> ----------------------------------------------------
>
>                 Key: HBASE-10309
>                 URL: https://issues.apache.org/jira/browse/HBASE-10309
>             Project: HBase
>          Issue Type: New Feature
>            Reporter: AcCud
>             Fix For: 0.94.17
>
>
> My use case: I have several tables where keys start with a timestamp. Because of this
and combined with the fact that I have set a 15 days retention period, after a period of time
results empty regions.
> I am sure that no write will occur in these region.
> It would be nice to have a tool to delete regions without being necessary to stop the
cluster.
> The easiest way for me is to have a tool that is able to delete all empty regions, but
there wouldn't be any problem to specify which region to delete.
> Something like:
> deleteRegion <tableName> <region>



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message