hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sergey Shelukhin (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-5258) Move coprocessors set out of RegionLoad
Date Thu, 08 Nov 2012 19:42:12 GMT

    [ https://issues.apache.org/jira/browse/HBASE-5258?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13493441#comment-13493441
] 

Sergey Shelukhin commented on HBASE-5258:
-----------------------------------------

There may be potential for using info about region server load that is more realtime/critical
than metrics channel, in master, for balancing/etc. Probably not important short term.
W.r.t. zk - why cannot master read zk nodes and apply rules too? ZK nodes may need to be made
smarter for this purpose. 
Assuming RSs will have to talk to ZK anyway, for large clusters it will avoid extra all-to-one
communication. Although master reading ZK to make decisions may introduce additional delay,
especially if master just watches the nodes and builds up internal state, instead of querying
entire state periodically.
Regardless, does having different coprocessors on the regions of the same table make sense?
If user has logic relying on coprocessors for some data querying or correctness it seems dangerous.


                
> Move coprocessors set out of RegionLoad
> ---------------------------------------
>
>                 Key: HBASE-5258
>                 URL: https://issues.apache.org/jira/browse/HBASE-5258
>             Project: HBase
>          Issue Type: Task
>            Reporter: Ted Yu
>            Priority: Critical
>
> When I worked on HBASE-5256, I revisited the code related to Ser/De of coprocessors set
in RegionLoad.
> I think the rationale for embedding coprocessors set is for maximum flexibility where
each region can load different coprocessors.
> This flexibility is causing extra cost in the region server to Master communication and
increasing the footprint of Master heap.
> Would HServerLoad be a better place for this set ?
> If required, region server should calculate disparity of loaded coprocessors among regions
and send report through HServerLoad

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message