[
https://issues.apache.org/jira/browse/HBASE-1502?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13009368#comment-13009368
]
stack commented on HBASE-1502:
------------------------------
The Master needs to know cluster load for load balancing and so we can display
load aggregate in the UI. Currently this is carried on the back of the
heatbeat as a serialized HServerLoad. The intent was to put this loading
elsewhere than up on the heartbeat. Thought was to put it out in zk.
At the hackathon, it was thought that on a cluster of any reasonable size, the
number of writes would be too high (Think 700 node cluster writing load every
1-5 seconds). Todd raised the issue after talking with Patrick; zk blocks
reads while taking on writes so our read rate would go down.
Chatting more, the thought was to just leave up the heartbeat connection --
call it something else though (jgray) -- but remove HMsg and just pass HSLs.
We talked of master querying on a period all RSs but for now, we have something
that works so just keep it up it was thought.
> Remove need for heartbeats in HBase
> -----------------------------------
>
> Key: HBASE-1502
> URL: https://issues.apache.org/jira/browse/HBASE-1502
> Project: HBase
> Issue Type: Task
> Reporter: Nitay Joffe
> Assignee: stack
> Priority: Blocker
> Fix For: 0.92.0
>
> Attachments: 1502-4.txt, 1502-v2.txt, 1502-v5.txt, 1502-v6.txt,
> 1502.txt
>
>
> HBase currently uses heartbeats between region servers and the master,
> piggybacking information on them when it can. This issue is to investigate if
> we can get rid of the need for those using ZooKeeper events.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
