[ 
https://issues.apache.org/jira/browse/AMBARI-13109?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16120333#comment-16120333
 ] 

Ted Yu commented on AMBARI-13109:
---------------------------------

The max logs parameter plays an important role in preventing fragmentation in 
memstore.

> Verify that max logs parameter wouldn't cause premature region flush
> --------------------------------------------------------------------
>
>                 Key: AMBARI-13109
>                 URL: https://issues.apache.org/jira/browse/AMBARI-13109
>             Project: Ambari
>          Issue Type: Improvement
>            Reporter: Ted Yu
>
> For hbase, the following relationship should be maintained:
> dfs.blocksize * 0.95 *  hbase.regionserver.maxlogs (default 32) > 
> hbase.regionserver.global.memstore.upperLimit * HBASE_HEAPSIZE
> The rationale is to avoid premature region flush due to log roll which is 
> triggered by insufficiently high max logs.
> When user changes the values for max logs, memstore upper limit or max heap 
> size, we should verify that the above relationship is satisfied.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to