[
https://issues.apache.org/jira/browse/HBASE-13008?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15065984#comment-15065984
]
wangyongqiang commented on HBASE-13008:
---------------------------------------
lower the value of hbase.regionserver.regionSplitLimit is too simple, may lead
to super-large region, leading to a slower read/write speed
i think we can use some reasonable methods to control and reduce the number
of regions
such as use region merge, create table region split control.
> Better default for hbase.regionserver.regionSplitLimit parameter.
> -----------------------------------------------------------------
>
> Key: HBASE-13008
> URL: https://issues.apache.org/jira/browse/HBASE-13008
> Project: HBase
> Issue Type: Improvement
> Reporter: Srikanth Srungarapu
> Assignee: Srikanth Srungarapu
> Priority: Minor
> Fix For: 1.0.0, 2.0.0, 1.1.0, 0.98.11
>
> Attachments: HBASE-13008.patch, HBASE-13008_v2.patch
>
>
> Currently, the default value for hbase.regionserver.regionSplitLimit is
> 2147483647. Couple of customers ran into trouble, as they ended up with too
> many regions(~7k) running on their region servers. Would it be possible for
> us to come up with better default or at least some new guard rail, which can
> be used for logging warnings when the number of regions are going overboard?
> Ref:
> [1]
> https://github.com/apache/hbase/blob/master/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/CompactSplitThread.java#L101
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
