You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hbase.apache.org by "Mikhail Antonov (JIRA)" <ji...@apache.org> on 2014/08/15 20:41:18 UTC

[jira] [Commented] (HBASE-11759) TableZnode should be cached

    [ https://issues.apache.org/jira/browse/HBASE-11759?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14098904#comment-14098904 ] 

Mikhail Antonov commented on HBASE-11759:
-----------------------------------------

There's another jira (linking) where table state is proposed to be moved out of ZK completely, to be permanently saves inside HTableDescriptor, and cached in memory inside TableStateManager. What do you think?

> TableZnode should be cached
> ---------------------------
>
>                 Key: HBASE-11759
>                 URL: https://issues.apache.org/jira/browse/HBASE-11759
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Virag Kothari
>            Assignee: Virag Kothari
>
> Observed on 0.98 with zk less assignment.
>  On receiving the rpc call about Region state transition, the rpc reader threads on master tries to figure whether the table is in disabling or disabled state. For this they all read tableznode which becomes a point of contention. We can just cache the state of table.



--
This message was sent by Atlassian JIRA
(v6.2#6252)