[jira] Commented: (HBASE-1030) Bit of polish on HBASE-1018

["stack (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-1030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12651188#action_12651188 ] 

stack commented on HBASE-1030:
------------------------------

Remove log of the serverload on regionserver side; will get real fat when loads of regions.

> Bit of polish on HBASE-1018
> ---------------------------
>
>                 Key: HBASE-1030
>                 URL: https://issues.apache.org/jira/browse/HBASE-1030
>             Project: Hadoop HBase
>          Issue Type: Bug
>            Reporter: stack
>             Fix For: 0.19.0
>
>
> HBASE-1018 added new attributes to serverload and then in the master UI, outputted the new detailed serverload.  Looking at this w/ jgray, if hundreds of regions per server, the UI will blow out trying to list per region load.  Jgray suggested that region-level detail instead should show over on the regionserver itself.  Will change the UI to do as he suggests and just leave the new summary memory used, etc., showing in the load box over in master UI.
> Doing math, we at first were afraid that the new fatter HServerLoad instance would overload the master when hundreds of regions but doing the math and counting the attributes that make up the new HSL, it looks like about ~128 characters per server every 3 second heartbeat, not too much.. so that should be fine.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.